AI Governance Initiative Cites Scott M. Graffius Research on AI Hallucinations

05 June 2026

BY SCOTT M. GRAFFIUS | ScottGraffius.com

ai-governance-initiative-uses-graffius-research---header-lwres

The Problem That Won't Go Away

Organizations deploying generative AI encounter the jarring problem of incorrect outputs. Not vague, not incomplete — wrong. A fabricated citation. A materially misquoted regulation. A fictitious financial figure. These are AI hallucinations — one of the most persistent and consequential issues in modern AI.

AI hallucinations can cause reputational harm, erosion of trust, and other serious consequences for organizations. For publicly traded companies and those in regulated industries — such as healthcare, finance, government — the risks also include compliance breaches.

Research That's Driving the Governance Conversation

That is why the question Scott M. Graffius set out to answer matters so much.

Graffius helps organizations unlock breakthrough performance and generate billions in business value through leadership in AI, Agile, and project, program, portfolio, and enterprise PMO management. He is a highly cited authority across those areas, which includes AI, the "exotic team dynamics" that emerge when advanced artificial intelligence collaborates as a teammate, and more.

One question has garnered significant attention:

Are AI hallucinations improving or worsening?

Graffius' article, "Are AI Hallucinations Getting Better or Worse? We Analyzed the Data," takes a rigorous, objective, evidence-based look at hallucination rates and trends. It moves beyond anecdotes and assumptions to examine what the data show.

The answer, it turns out, is more nuanced than optimists or skeptics tend to acknowledge. And that nuance — that hallucinations are declining on constrained, grounded tasks while surging on complex reasoning and open-ended factual recall — is exactly what has made the research valuable to governance practitioners endeavoring to establish reliable AI solutions.

Graffius' respective work has since been cited across multiple publications. Select examples follow. The References section of this article provides details, including links to each publication.

Agenda Digitale.
AIUnpacking.
Asky.
BuzzStream.
COMPAiSS. (Two publications.)
Cottrill Research.
Firecrawl.
FlowEdge.
Grokipedia.
Laughman AI.
LoudScale.
LSS4Managers.
Mental Momentum AI.
Perplexity.
Scintilena.
Seopedia.
Swizero.
And more.

The list includes two publications from the Canadian AI governance initiative, COMPAiSS. Those are noteworthy because they reflect growing attention to AI accountability, reliability, and governance alongside AI capability.

COMPAiSS: Building the Case for Governed AI

COMPAiSS is an AI governance initiative built specifically for environments where errors carry real consequences. Its focus areas — higher education, healthcare, government, and professional organizations — share a common thread: the cost of an AI mistake is not just inconvenient, it is potentially harmful, liable, or irreversible.

Its framework rests on the premise that AI systems deployed in regulated settings must prioritize controlled, auditable behavior over unconstrained generation. Where most AI discourse centers on capability (what the model can do), COMPAiSS redirects attention to accountability (what the model should do), and what happens when it doesn't.

That means a strong emphasis on reliability, traceability, bounded outputs, and clear accountability in AI-assisted decision-making. It is a framework designed not for experimentation, but for institutional deployment where predictability is non-negotiable.

Among its published work is "Hallucinations by Design: A Cross-Model Assessment of How Major AI Consulting Firms Advise Clients to Manage Generative AI Errors," which analyzes how major organizations — including Deloitte, PwC, KPMG, and Microsoft — approach hallucination mitigation in practice. Graffius' research on hallucination trends serves as a cited foundation in that analysis, grounding its findings in independent, empirical data. A companion publication, "Breaking Out of the Matrix," extends that work into the governance constraints of AI systems and the risks posed by unconstrained generative models in institutional settings — and draws on Graffius' work there as well.

Together, these publications reflect COMPAiSS' conviction that controlled inference and auditability are not optional features — they are prerequisites for responsible AI in high-stakes environments.

Why Independent Research Shapes Governance Frameworks

Governance frameworks do not emerge from a vacuum. They are built on evidence: on empirical research that documents the actual performance of AI systems, including where they fail, and how those failures change over time. Independent researchers who investigate these questions provide the kind of foundational insight that institutional frameworks need to be credible. That is the gap that work like Graffius' helps fill.

Conclusion: Getting AI Right Requires More Than Getting It Working

The conversation around AI hallucinations is ongoing and intensifying as deployments deepen and the margin for error narrows. What's becoming clear is that responsible AI adoption requires more than technical capability. It requires governance infrastructure, accountability mechanisms, and a shared body of research to draw from.

This also points to a broader shift underway in enterprise AI: away from unconstrained generative models deployed on enthusiasm, and toward governed, auditable systems built for institutional accountability. The organizations leading that shift are not simply adopting AI — they are architecting the conditions under which AI can be trusted, drawing on the collective work of researchers, practitioners, and governance architects. The inclusion of Graffius' research in COMPAiSS reflects that broader effort to build AI systems that organizations can trust.

AI's promise gets it in the door. Quality, reliability, and governance are what keep it there.

ai-governance-initiative-uses-graffius-research---quote-IG-v2-lwres

ai-governance-initiative-uses-graffius-research---quote-IG-v2-lwres

Graffius has generated over $2.51 billion in business value for Fortune 500 companies and other organizations around the world. Put that track record to work for you. For speaking engagements, use the request form; for other inquiries, email him.

References

Agenda Digitale. (2026, May 6). L’AI può aiutare a studiare, ma solo se insegna a ragionare. https://www.agendadigitale.eu/scuola-digitale/lai-puo-aiutare-a-studiare-ma-solo-se-insegna-a-ragionare/
AIUnpacking. (2026, April 15). How to Avoid AI Hallucinations: A Practical 2026 Checklist. https://aiunpacking.com/blog/how-to-avoid-ai-hallucinations/
Asky. (2026, May 26). Navigating AI perception challenges in social marketing. https://askylabs.com/learn/social-and-brand-signal-amplification/ai-perception-challenges-social-marketing
BuzzStream. (2026, April 8). Do Americans Use AI for News? https://www.buzzstream.com/blog/ai-news-usage/
COMPAiSS. (n.d.). Hallucination rates [PDF]. https://compaiss.ca/assets/hallucination_rates-fr.pdf
COMPAiSS. (2026, May). Breaking out of the matrix (Version 3.2) [PDF]. https://compaiss.ca/assets/breaking_out_of_the_matrix_v3_2.pdf
Cottrill Research. (n.d.). AI Hallucination Benchmark Resources Added to Index Collection. https://cottrillresearch.com/ai-hallucination-benchmark-resources-added-to-index-collection/
Firecrawl. (2026, April 28). Training Data vs. Retrieved Data vs. Live Web Data: What Data Makes Your AI Agent Smarter. https://www.firecrawl.dev/blog/training-vs-retrieved-vs-live-web-data
FlowEdge. (2026, March 13). AI Tools: ChatGPT Limitations as an AI Proposal Writer. https://getflowedge.com/ai-tools/chatgpt-limitations-as-an-ai-proposal-writer/
Graffius, S. M. (2025, June 17). PMI’s Infinity AI Gets the Basics of Team Development Alarmingly and Repeatedly Wrong. https://doi.org/10.13140/RG.2.2.32957.93927
Graffius, S. M. (2026, January 7). Are AI Hallucinations Getting Better or Worse? We Analyzed the Data. https://doi.org/10.13140/RG.2.2.33179.53285 [COMPAiSS lists it in French as: Graffius S.M. (2026). Les hallucinations de l'IA s'améliorent-elles ou s'aggravent-elles? Nous avons analysé les données. DOI : 10.13140/RG.2.2.33179.53285.]
Grokipedia. (n.d.). ChatGPT. https://grokipedia.com/page/ChatGPT
Ha, D. (2026, May 2). AI Hallucinations Risk in LLM Lifecycle [LinkedIn post]. https://www.linkedin.com/posts/dannyhaprofile_aihallucinations-aigovernance-llmrisk-activity-7456485913309810688-bfcI/
Laughman AI. (2026, May 16). 知能のブリーダー：人間とAIの存在論的境界、そして自律進化するデジタル生態系の夜明け. https://note.com/betaitohuman/n/n32c36b9890c2
LoudScale. (2026, February 15). How to Fact-Check AI Content Before Publishing (2026). https://loudscale.com/blog/fact-check-ai-content-before-publishing/
LSS4Managers. (2026, February 27). How to Detect AI Hallucinations? A Way to Deal with Fake Data! https://lss4managers.com/jak-wykryc-halucynacje-ai-sposob-na-falszywe-dane/
Mental Momentum AI. (n.d.). How Could AI Change Scientific Discovery? 5 Scenarios. https://research.mental-momentum.ai/r/how-ai-could-change-scientific-discovery-rh80aj
Perplexity. (2026, February 12). Study Finds Celebrated AI Model Guesses Answers, Lacks Cognition. Perplexity AI stand-alone article. https://www.perplexity.ai/page/experts-say-ai-lacks-true-huma-JfKEeQvcRhCddq0tgTAiMw
Scintilena. (2026, May 1). Perché le AI Generative Falliscono in Geografia: Errori di Localizzazione, Prossimità e Allucinazioni. https://www.scintilena.com/perche-le-ai-generative-falliscono-in-geografia-errori-di-localizzazione-prossimita-e-allucinazioni/05/01/
Seopedia. (n.d.). Que es un RAG en IA (Retrieval-Augmented Generation). https://seopedia.es/blog/que-es-un-rag-en-ia/
Swizero. (2026, March 28). Your AI Email Assistant Saves 40% But 77% Feel Busier. https://www.swizero.com/blog/how-ai-email-assistants-actually-work
Walczak, T. (2026, February 26). Jak Wykryć Halucynacje AI? Sposób na Fałszywe Dane! [LinkedIn post]. https://www.linkedin.com/pulse/jak-wykry%C4%87-halucynacje-ai-spos%C3%B3b-na-fa%C5%82szywe-dane-walczak--ynluf/

About COMPAiSS

COMPAiSS is a Canada-based AI governance initiative focused on institutional environments where accuracy, accountability, and policy fidelity are non-negotiable. Rather than treating AI as a general-purpose generative system, COMPAiSS is designed for regulated settings such as universities, healthcare systems, government agencies, and professional bodies, where information must be delivered strictly within authorized institutional boundaries. Its architecture emphasizes execution-gated inference — meaning AI responses are only generated when a question falls within defined scope and when authoritative institutional sources are available to support an answer. In contrast to conventional “generate-then-filter” AI systems, COMPAiSS positions authorization as a prerequisite to inference, aiming to reduce hallucinations by preventing unsupported generation at the source. The initiative’s work is positioned within a broader effort to reframe AI not just as a productivity tool, but as a governed institutional system aligned with compliance, auditability, and operational risk management.

About Scott M. Graffius

scott_m_graffius_-_blue_-_1000x1000_-lwres

Scott M. Graffius is a strategic transformation leader who drives AI, Agile, and broader business and technology initiatives to deliver measurable value across projects, programs, portfolios, and PMOs. He is an expert in the teamwork tradecraft of both human and human-AI teams, including the “exotic team dynamics” that emerge. He is also an authority on the temporal patterns of social media, including the half-life of audience engagement.

He’s a practitioner, researcher, thought leader, award-winning author, and keynote speaker who’s taken the stage at 98 conferences and other events across 25 countries.

He’s delivered over $2.51 billion in value for Fortune 500 companies and other leaders in technology, entertainment, financial services, healthcare, and beyond.

Businesses, professional associations, government agencies, and universities use Graffius and feature his work. Examples include Adobe, Bayer, Boston University, Ford, Gartner, Harvard Medical School, IEEE, Johns Hopkins University, Microsoft, MSN, National Academy of Sciences, Oracle, Pinterest Inc., Project Management Institute, UC San Diego, Verizon, Yale University, and others.

The following sections provide additional information on his experience, contributions, and influence.

Experience

Graffius heads the professional services firm Exceptional PPM and PMO Solutions, along with its subsidiary Exceptional Agility. These consultancies offer strategic and tactical advisory, training, embedded expertise, and consulting services to the public, private, and government sectors. They help organizations enhance their capabilities and results in agile, project management, program management, portfolio management, and PMO leadership, supporting innovation and driving competitive advantage. The consultancies confidently back services with a Delighted Client Guarantee™.

Graffius is a former VP of project management with a publicly traded provider of diverse consumer products and services over the Internet. Before that, he ran and supervised the delivery of projects and programs in public and private organizations with businesses ranging from e-commerce to advanced technology products and services, retail, manufacturing, entertainment, and more.

He has experience with consumer, business, reseller, government, and international markets.

Award-Winning Author

Graffius has authored three books.

Agile Scrum: Your Quick Start Guide with Step-by-Step Instructions, his first book, earned 17 awards.
Agile Transformation: A Brief Story of How an Entertainment Company Developed New Capabilities and Unlocked Business Agility to Thrive in an Era of Rapid Change, his second book, was named one of the best Scrum books of all time by BookAuthority.
Agile Protocol: The Transformation Ultimatum, his third book and his first work of fiction, was released in April 2025. The book trailer is on YouTube.

International Public Speaker

Organizations worldwide engage Graffius to present on tech (including AI), Agile, project management, program management, portfolio management, and PMO leadership. He crafts and delivers unique and compelling talks and workshops. Graffius has conducted 98 sessions across 25 countries. Select examples of events include Agile Trends Gov, BSides (Newcastle Upon Tyne), Conf42 Quantum Computing, DevDays Europe, DevOps Institute, DevOpsDays (Geneva), Frug’Agile, IEEE, Microsoft, Scottish Summit, Scrum Alliance RSG (Nepal), Techstars, and W Love Games International Video Game Development Conference (Helsinki), and more.

With an average rating of 4.81 (on a scale of 1-5), sessions are highly valued.

The speaker engagement request form is here.

Thought Leadership and Influence

Prominent businesses, professional associations, government agencies, and universities have showcased Graffius and his contributions—spanning his books, talks, workshops, and beyond. Select examples include:

Adobe,
American Management Association,
Amsterdam Public Health Research Institute,
Bayer,
BMC Software,
Boston University,
Broadcom,
Cisco,
Coburg University of Applied Sciences and Arts - Germany,
Computer Weekly,
Constructor University - Germany,
Data Governance Success,
Deimos Aerospace,
DevOps Institute,
Dropbox,
EU's European Commission,
Ford Motor Company,
Gartner,
GoDaddy,
Harvard Medical School,
Hasso Plattner Institute - Germany,
IEEE,
Innovation Project Management,
Johns Hopkins University,
Journal of Neurosurgery,
Lam Research (Semiconductors),
Leadership Worthy,
Life Sciences Trainers and Educators Network,
London South Bank University,
Microsoft,
MSN,
NASSCOM,
National Academy of Sciences,
New Zealand Government,
Oracle,
Pinterest Inc.,
Project Management Institute,
Mary Raum (Professor of National Security Affairs, United States Naval War College),
SANS Institute,
SBG Neumark - Germany,
Singapore Institute of Technology,
Torrens University - Australia,
TBS Switzerland,
Tufts University,
UC San Diego,
UK Sports Institute,
University of Galway - Ireland,
US Department of Energy,
US National Park Service,
US Soccer,
US Tennis Association,
Verizon,
Wrike,
Yale University,
and many others.

Graffius has played a key role in the Project Management Institute (PMI) in developing professional standards. He was a member of multiple teams that authored, reviewed, and produced:

Agile Practice Guide – Second Edition
A Guide to the Project Management Body of Knowledge (PMBOK Guide) – Eighth Edition
A Guide to the Project Management Body of Knowledge (PMBOK Guide) – Sixth Edition
The Standard for Program Management – Fourth Edition
Practice Standard for Work Breakdown Structures – Second Edition
The Practice Standard for Project Estimating – Second Edition

He was also a subject matter expert reviewer of content for the PMI’s Congress. Beyond the PMI, Graffius also served as a member of the review team for two of the Scrum Alliance’s Global Scrum Gatherings.

Acclaimed Authority on Teamwork Tradecraft

Scott-M-Graffius-Phases-Of-Team-Development-2026-Update-v26010307G2-jpg-lwres

Graffius is a renowned authority on teamwork tradecraft. Informed by the research of Bruce W. Tuckman and Mary Ann C. Jensen, over 150 subsequent studies, and Graffius' first-hand professional experience with, and analysis of, team leadership and performance, Graffius created his "Phases of Team Development" intellectual property as a unique perspective and visual conveying the five phases of team development. First introduced in 2008 and periodically updated, his work provides a diagnostic and strategic guide for navigating team dynamics. It provides actionable insights for leaders across industries to develop high-performance teams. Its adoption by esteemed organizations such as Yale University, IEEE, Cisco, Microsoft, Ford, Oracle, Broadcom, the U.S. National Park Service, and the Journal of Neurosurgery, among others, highlights its utility and value, solidifying its status as an indispensable resource for elevating team performance and driving organizational excellence. In 2026, Graffius added human-AI teamwork—including the "exotic team dynamics" which emerge when advanced AI collaborates as a teammate—to his "Phases of Team Development."

The 2026 edition of Graffius' "Phases of Team Development" intellectual property is here.

Expert on Temporal Dynamics on Social Media Platforms

scott-m-graffius-lifespan-halflife-of-social-media-posts-update-for-2026-summary-visual-v26012207-jpg-lwres

Graffius is also an authority on temporal dynamics on social media platforms. His "Lifespan (Half-Life) of Social Media Posts" research—first published in 2018 and updated annually—delivers a precise quantitative analysis of post longevity across digital platforms, utilizing advanced statistical techniques to determine mean half-life with precision. It establishes a solid empirical base, effectively highlighting the ephemeral nature of content within social media ecosystems. Referenced and applied by leading entities—such as Fast Company, GoDaddy, Journal of Hand Surgery (European Volume), Ministère de la Culture (French Ministry of Culture), Pinterest Inc., PNAS, and Telecommunications Policy, among others—his research exemplifies methodological rigor and sustained significance in the field of digital informatics.

The 2026 edition of Graffius "Lifespan (Half-Life) of Social Media Posts" research is here.

Education and Professional Certifications

Graffius has a bachelor’s degree in psychology with a focus in Human Factors. He holds eight professional certifications:

Certified SAFe 6 Agilist (SA),
Certified Scrum Professional - ScrumMaster (CSP-SM),
Certified Scrum Professional - Product Owner (CSP-PO),
Certified ScrumMaster (CSM),
Certified Scrum Product Owner (CSPO),
Project Management Professional (PMP),
Lean Six Sigma Green Belt (LSSGB), and
IT Service Management Foundation (ITIL).

He is an active member of the Scrum Alliance, the Project Management Institute (PMI), and the Institute of Electrical and Electronics Engineers (IEEE).

Advancing AI, Agile, and Project/PMO Management

Scott M. Graffius continues to advance the fields of AI, Agile, and Project/PMO Management through his leadership, research, writing, and real-world impact. Businesses and other organizations leverage Graffius’ insights to drive their success.

Discover Scott’s Books

Connect with and follow Scott on LinkedIn, X, YouTube, Facebook, Threads, Bluesky, Mastodon, and ResearchGate.

pinterest-inc-references-scott-m-graffius0027-research---v1-3---lwres

cisco-features-scott-m.-graffius0027-0027phases-of-team-development0027-work---rectange---lwres

ieee-xplore-publication-featured-scott-m-graffius-phases-of-team-development-work---rev-sept-19-2024---hires

ucsd-featured-ip-of-scott-m-graffius---v25122907-giesel-lwres

johns-hopkins-university-features-work-of-scott-m-graffius---v24080107---tw---lwres

semiconductor-manufacturing-firm-lam-research-features-scott-m-graffius2019-2018phases-of-team-development2019-intellectual-property---tw-sz-format---lwres

bayer-licensed-ip-of-agile-expert-scott-m-graffius---agilescrumguide_com---lwres

ASG - Award Winning Agile Scrum Book by Scott M Graffius - v260427 - lwres

scott_m_graffius_agile_transformation_sq_lr_1000x1000

gifts-that-inspire-joy---scottgraffius-dot-com---v251014---lwres

Sign up for Miro—it's free!

(Want more features? You can always upgrade to a paid plan.)

scottgraffius-com-article-sections-v2026-more-lwres

List of Additional Articles

scottgraffius-com-article-sections-v2026-cite-lwres

How to Cite This Article

Graffius, S. M. (2026, June 5). AI Governance Initiative Cites Scott M. Graffius Research on AI Hallucinations. ScottGraffius.com. https://scottgraffius.com/blog/files/ai-governance-initiative-uses-graffius-research.html

scottgraffius-com-article-sections-v2026-doi-lwres

Digital Object Identifier (DOI)

Coming soon

scottgraffius-com-article-sections-v2026-ackn-lwres

Content Acknowledgements

Names, marks, and content are the property of their respective owners.

scottgraffius-com-article-sections-v2026-tags-hashtags-lwres

#Agile
#AI
#AIGovernance
#AIHallucinations
#AIReliability
#AISafety
#ArtificialIntelligence
#Auditability
#COMPAiSS
#Compliance
#DigitalTransformation
#EmergingTech
#EnterpriseAI
#GenerativeAI
#Governance
#LLM
#MachineLearning
#ModelRisk
#PMO
#ProjectManagement
#ResponsibleAI
#RiskManagement
#TechGovernance
#TrustInAI
#TrustworthyAI

scottgraffius-com-article-sections-v2026-notes-lwres

Post-Publication Notes

If there are any supplements or updates to this article after the date of publication, they will appear here.

Copyright

Copyright © Scott M. Graffius. All rights reserved.

Content on this site—including text, images, videos, and data—may not be used for training or input into any artificial intelligence, machine learning, or automatized learning systems, or published, broadcast, rewritten, or redistributed without the express written permission of Scott M. Graffius.

Tags: Agile • AI • AI Governance • AI Hallucinations • AI Reliability • COMPAiSS • Enterprise AI • Generative AI • Responsible AI • Scott M. Graffius • Technology • Do AI Hallucinations Improve or Worsen Over Time? We Analyzed the Data • Project Management • Machine Learning • Institutional Accountability • Governance Frameworks

Scott M. Graffius Contributed to The Standard for Artificial Intelligence in Portfolio, Program, and Project Management

09 June 2026

BY SCOTT M. GRAFFIUS | ScottGraffius.com

scott-m-graffius-contributed-to-pmi-s𝘁𝗮𝗻𝗱𝗮𝗿𝗱-𝗳𝗼𝗿-ai-i𝗻-pppm---lwres

Recommended Citation

Graffius, S. M. (2026, June 9). Scott M. Graffius Contributed to The Standard for Artificial Intelligence in Portfolio, Program, and Project Management. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.10979.52003

Artificial intelligence is reshaping how organizations plan, execute, and deliver value across portfolios, programs, and projects. Recognizing this shift, the Project Management Institute (PMI) developed The Standard for Artificial Intelligence in Portfolio, Program, and Project Management, a landmark publication that gives practitioners a principled, structured, and globally recognized framework for navigating AI in the complexity of modern delivery environments.

Scott M. Graffius contributed to this standard, as shown in the visual at the top of this article and below.

About The Standard for Artificial Intelligence in Portfolio, Program, and Project Management

The increasing adoption of AI is transforming the landscape of portfolio, program, and project management (PPPM). As AI continues to evolve, it offers significant opportunities to enhance decision-making, optimize resource allocation, and provide valuable insights. To support PPPM professionals in navigating this complex and rapidly changing environment, PMI developed The Standard for Artificial Intelligence in Portfolio, Program, and Project Management.

This standard is built on the solid foundation of clear principles established by PMI, which help guide and inform behaviors, as well as performance domains that focus on critical activities and functions for effective delivery and implementation. Through this guidance, organizations can help ensure their AI initiatives are implemented in a way that is transparent, fair, and compliant with applicable regulations. The details provided are relevant across various industries and intended to support organizations of all sizes in achieving their AI-driven objectives.

Key components of the standard include:

A comprehensive framework for the effective and responsible integration of AI into PPPM practices.
A human-centered approach, ensuring that AI insights are aligned with human oversight.
Eight guiding principles and five critical performance domains for using AI in PPPM.
A proposed life cycle for AI implementations, along with tailoring guidance.
Examination of the ethical and legal considerations for AI in PPPM.

By adopting this standard, PPPM professionals can enhance their practices, stay at the forefront of AI adoption, and ensure that their organizations maximize the benefits of AI while maintaining accountability and delivering value to stakeholders.

The standard is aligned with other PMI standards and underwent a rigorous review process. It is approved by the American National Standards Institute (ANSI) — a distinction that signals the document's rigor, credibility, and suitability for real-world professional application at the highest level.

PMI members can download the PDF for free as part of their membership benefits here. The publication is also available in print and ebook formats from Amazon here, as well as other sellers.

This is a publication of consequence. The PMI has been setting the gold standard in project management for over 50 years; and with this standard, it addresses one of the most significant transformations the profession has faced. The standard reflects PMI's long-standing approach: convening global communities of experienced practitioners, subjecting draft content to rigorous review, and producing guidance that is both principled and practically applicable.

Scott M. Graffius' Contribution

Scott M. Graffius contributed to The Standard for Artificial Intelligence in Portfolio, Program, and Project Management. His involvement reflects his sustained commitment to advancing the profession at its highest levels — and his particular expertise at the intersection of AI and delivery management.

Graffius has now contributed to seven PMI industry standards and guidance publications:

The Standard for Artificial Intelligence in Portfolio, Program, and Project Management
Agile Practice Guide – Second Edition
A Guide to the Project Management Body of Knowledge (PMBOK Guide) – Eighth Edition
A Guide to the Project Management Body of Knowledge (PMBOK Guide) – Sixth Edition
The Standard for Program Management – Fourth Edition
Practice Standard for Work Breakdown Structures – Second Edition
The Practice Standard for Project Estimating – Second Edition

This body of contribution is exceptional. PMI standards are developed through a volunteer-driven, consensus-based process in which expert practitioners participate. Contributing to one PMI standard is an achievement. Contributing to seven reflects a sustained commitment to advancing the profession across methodologies, disciplines, and generations of practice.

It is also worth noting the relevance of this newest contribution to Graffius' broader research and thought leadership. He is a recognized authority on Agile, Project Management, and AI – including human-AI collaboration, and "exotic team dynamics" that emerge when advanced AI systems collaborate as teammates, AI hallucinations, governance, and more. His respective work has been cited businesses, professional associations, government agencies, universities, publications, and media outlets around the world. Contributing to The Standard for Artificial Intelligence in Portfolio, Program, and Project Management — a standard that addresses the responsible integration of AI into organizational delivery practices — aligns with his body of work.

scott-m-graffius-contributed-to-pmi-s𝘁𝗮𝗻𝗱𝗮𝗿𝗱-𝗳𝗼𝗿-ai-i𝗻-pppm---IG-lwres

Graffius has generated over $2.51 billion in business value for Fortune 500 companies and other organizations around the world. Put that track record to work for you. For speaking engagements, use the request form; for other inquiries, email him.

About the Project Management Institute

The Project Management Institute (PMI) is the leading global association for project management professionals. Founded in 1969, PMI advances the profession through research, standards development, education, and certification. Its Project Management Professional (PMP) credential is recognized worldwide as the benchmark for project management expertise. Through its global network, PMI connects practitioners, elevates shared knowledge, and champions the strategic value of effective project leadership across industries. Visit the PMI website at https://www.pmi.org to learn more.

Bibliography/References

Graffius, S. M. (2025, February 12). Scott M. Graffius' Role in Advancing Project Management Institute (PMI) Standards Excellence. https://scottgraffius.com/blog/files/scott-m-graffius-role-in-advancing-project-management-institute-(pmi)-standards.html

Graffius, S. M. (2025, November 13). Scott M. Graffius Contributed to and Reviewed the PMBOK Guide, 8th Edition — the Global Standard for Project Management. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.25557.69601

Graffius, S. M. (2026, January 3). Scott M. Graffius' Phases of Team Development - Applied to Human Teams and Human-AI Teams: 2026 Update. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.18184.89601

Graffius, S. M. (2026, January 7). Are AI Hallucinations Getting Better or Worse? We Analyzed the Data. https://doi.org/10.13140/RG.2.2.33179.53285

Graffius, S. M. (2026, June 3). Scott M. Graffius Contributed to the Agile Practice Guide - Second Edition. ScottGraffius.com. https://scottgraffius.com/blog/files/scott-m-graffius-contributed-to-agile-practice-guide-second-edition.html

Graffius, S. M. (2026, June 5). AI Governance Initiative Cites Scott M. Graffius Research on AI Hallucinations. ScottGraffius.com. https://scottgraffius.com/blog/files/ai-governance-initiative-uses-graffius-research.html

Project Management Institute. (n.d.). https://www.pmi.org

Project Management Institute. (2006). Practice standard for work breakdown structures (2nd ed.). Project Management Institute.

Project Management Institute. (2017). The standard for program management (4th ed.). Project Management Institute.

Project Management Institute. (2017). A guide to the project management body of knowledge (PMBOK guide) (6th ed.). Project Management Institute.

Project Management Institute. (2019). The practice standard for project estimating (2nd ed.). Project Management Institute.

Project Management Institute. (2026). A guide to the project management body of knowledge (PMBOK guide) (8th ed.). Project Management Institute.

Project Management Institute. (2026). Agile practice guide (2nd ed.). Project Management Institute.

Project Management Institute. (2026). The Standard for Artificial Intelligence in Portfolio, Program, and Project Management. Project Management Institute.

Google. (n.d.). Google Books.

About Scott M. Graffius

Scott M. Graffius is a strategic transformation leader who drives AI, Agile, and broader business and technology initiatives to deliver measurable value across projects, programs, portfolios, and PMOs. He is an expert in the teamwork tradecraft of both human and human-AI teams, including the “exotic team dynamics” that emerge. He is also an authority on the temporal patterns of social media, including the half-life of audience engagement.

He’s a practitioner, researcher, thought leader, award-winning author, and keynote speaker who’s taken the stage at 98 conferences and other events across 25 countries.

He’s delivered over $2.51 billion in value for Fortune 500 companies and other leaders in technology, entertainment, financial services, healthcare, and beyond.

Businesses, professional associations, government agencies, and universities use Graffius and feature his work. Examples include Adobe, Bayer, Boston University, Ford, Gartner, Harvard Medical School, IEEE, Johns Hopkins University, Microsoft, MSN, National Academy of Sciences, Oracle, Pinterest Inc., Project Management Institute, UC San Diego, Verizon, Yale University, and others.

The following sections provide additional information on his experience, contributions, and influence.

Experience

Graffius heads the professional services firm Exceptional PPM and PMO Solutions, along with its subsidiary Exceptional Agility. These consultancies offer strategic and tactical advisory, training, embedded expertise, and consulting services to the public, private, and government sectors. They help organizations enhance their capabilities and results in agile, project management, program management, portfolio management, and PMO leadership, supporting innovation and driving competitive advantage. The consultancies confidently back services with a Delighted Client Guarantee™.

Graffius is a former VP of project management with a publicly traded provider of diverse consumer products and services over the Internet. Before that, he ran and supervised the delivery of projects and programs in public and private organizations with businesses ranging from e-commerce to advanced technology products and services, retail, manufacturing, entertainment, and more.

He has experience with consumer, business, reseller, government, and international markets.

Award-Winning Author

Graffius has authored three books.

Agile Scrum: Your Quick Start Guide with Step-by-Step Instructions, his first book, earned 17 awards.
Agile Transformation: A Brief Story of How an Entertainment Company Developed New Capabilities and Unlocked Business Agility to Thrive in an Era of Rapid Change, his second book, was named one of the best Scrum books of all time by BookAuthority.
Agile Protocol: The Transformation Ultimatum, his third book and his first work of fiction, was released in April 2025. The book trailer is on YouTube.

International Public Speaker

Organizations worldwide engage Graffius to present on tech (including AI), Agile, project management, program management, portfolio management, and PMO leadership. He crafts and delivers unique and compelling talks and workshops. Graffius has conducted 98 sessions across 25 countries. Select examples of events include Agile Trends Gov, BSides (Newcastle Upon Tyne), Conf42 Quantum Computing, DevDays Europe, DevOps Institute, DevOpsDays (Geneva), Frug’Agile, IEEE, Microsoft, Scottish Summit, Scrum Alliance RSG (Nepal), Techstars, and W Love Games International Video Game Development Conference (Helsinki), and more.

With an average rating of 4.81 (on a scale of 1-5), sessions are highly valued.

The speaker engagement request form is here.

Thought Leadership and Influence

Prominent businesses, professional associations, government agencies, and universities have showcased Graffius and his contributions—spanning his books, talks, workshops, and beyond. Select examples include:

Adobe,
American Management Association,
Amsterdam Public Health Research Institute,
Bayer,
BMC Software,
Boston University,
Broadcom,
Cisco,
Coburg University of Applied Sciences and Arts - Germany,
Computer Weekly,
Constructor University - Germany,
Data Governance Success,
Deimos Aerospace,
DevOps Institute,
Dropbox,
EU's European Commission,
Ford Motor Company,
Gartner,
GoDaddy,
Harvard Medical School,
Hasso Plattner Institute - Germany,
IEEE,
Innovation Project Management,
Johns Hopkins University,
Journal of Neurosurgery,
Lam Research (Semiconductors),
Leadership Worthy,
Life Sciences Trainers and Educators Network,
London South Bank University,
Microsoft,
MSN,
NASSCOM,
National Academy of Sciences,
New Zealand Government,
Oracle,
Pinterest Inc.,
Project Management Institute,
Mary Raum (Professor of National Security Affairs, United States Naval War College),
SANS Institute,
SBG Neumark - Germany,
Singapore Institute of Technology,
Torrens University - Australia,
TBS Switzerland,
Tufts University,
UC San Diego,
UK Sports Institute,
University of Galway - Ireland,
US Department of Energy,
US National Park Service,
US Soccer,
US Tennis Association,
Verizon,
Wrike,
Yale University,
and many others.

Graffius has played a key role in the Project Management Institute (PMI) in developing professional standards. He was a member of multiple teams that authored, reviewed, and produced:

The Standard for Artificial Intelligence in Portfolio, Program, and Project Management
Agile Practice Guide – Second Edition
A Guide to the Project Management Body of Knowledge (PMBOK Guide) – Eighth Edition
A Guide to the Project Management Body of Knowledge (PMBOK Guide) – Sixth Edition
The Standard for Program Management – Fourth Edition
Practice Standard for Work Breakdown Structures – Second Edition
The Practice Standard for Project Estimating – Second Edition

He was also a subject matter expert reviewer of content for the PMI’s Congress. Beyond the PMI, Graffius also served as a member of the review team for two of the Scrum Alliance’s Global Scrum Gatherings.

Acclaimed Authority on Teamwork Tradecraft

Graffius is a renowned authority on teamwork tradecraft. Informed by the research of Bruce W. Tuckman and Mary Ann C. Jensen, over 150 subsequent studies, and Graffius' first-hand professional experience with, and analysis of, team leadership and performance, Graffius created his "Phases of Team Development" intellectual property as a unique perspective and visual conveying the five phases of team development. First introduced in 2008 and periodically updated, his work provides a diagnostic and strategic guide for navigating team dynamics. It provides actionable insights for leaders across industries to develop high-performance teams. Its adoption by esteemed organizations such as Yale University, IEEE, Cisco, Microsoft, Ford, Oracle, Broadcom, the U.S. National Park Service, and the Journal of Neurosurgery, among others, highlights its utility and value, solidifying its status as an indispensable resource for elevating team performance and driving organizational excellence. In 2026, Graffius added human-AI teamwork—including the "exotic team dynamics" which emerge when advanced AI collaborates as a teammate—to his "Phases of Team Development."

The 2026 edition of Graffius' "Phases of Team Development" intellectual property is here.

Expert on Temporal Dynamics on Social Media Platforms

Graffius is also an authority on temporal dynamics on social media platforms. His "Lifespan (Half-Life) of Social Media Posts" research—first published in 2018 and updated annually—delivers a precise quantitative analysis of post longevity across digital platforms, utilizing advanced statistical techniques to determine mean half-life with precision. It establishes a solid empirical base, effectively highlighting the ephemeral nature of content within social media ecosystems. Referenced and applied by leading entities—such as Fast Company, GoDaddy, Journal of Hand Surgery (European Volume), Ministère de la Culture (French Ministry of Culture), Pinterest Inc., PNAS, and Telecommunications Policy, among others—his research exemplifies methodological rigor and sustained significance in the field of digital informatics.

The 2026 edition of Graffius "Lifespan (Half-Life) of Social Media Posts" research is here.

Education and Professional Certifications

Graffius has a bachelor’s degree in psychology with a focus in Human Factors. He holds eight professional certifications:

Certified SAFe 6 Agilist (SA),
Certified Scrum Professional - ScrumMaster (CSP-SM),
Certified Scrum Professional - Product Owner (CSP-PO),
Certified ScrumMaster (CSM),
Certified Scrum Product Owner (CSPO),
Project Management Professional (PMP),
Lean Six Sigma Green Belt (LSSGB), and
IT Service Management Foundation (ITIL).

He is an active member of the Scrum Alliance, the Project Management Institute (PMI), and the Institute of Electrical and Electronics Engineers (IEEE).

Advancing AI, Agile, and Project/PMO Management

Scott M. Graffius continues to advance the fields of AI, Agile, and Project/PMO Management through his leadership, research, writing, and real-world impact. Businesses and other organizations leverage Graffius’ insights to drive their success.

Discover Scott’s Books

Connect with and follow Scott on LinkedIn, X, YouTube, Facebook, Threads, Bluesky, Mastodon, and ResearchGate.

Sign up for Miro—it's free!

(Want more features? You can always upgrade to a paid plan.)

scottgraffius-com-article-sections-v2026-more-lwres

List of Additional Articles

How to Cite This Article

Graffius, S. M. (2026, June 9). Scott M. Graffius Contributed to The Standard for Artificial Intelligence in Portfolio, Program, and Project Management. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.10979.52003

Digital Object Identifier (DOI)

DOI: 10.13140/RG.2.2.10979.52003

Content Acknowledgements

Project Management Institute (PMI) name, mark, and content are the property of PMI.

Graffius’ article includes limited excerpts of the PMI publication, attributed and used under fair use doctrine for purposes of news reporting, commentary, and analysis.

Graffius’ article copyright © Scott M. Graffius.

#ArtificialIntelligence
#AIGovernance
#ProjectManagement
#ProjectManagementInstitute
#PortfolioManagement
#ProgramManagement
#Leadership
#ANSIStandards

Post-Publication Notes

If there are any supplements or updates to this article after the date of publication, they will appear here.

Copyright

Copyright © Scott M. Graffius. All rights reserved.

Content on this site—including text, images, videos, and data—may not be used for training or input into any artificial intelligence, machine learning, or automatized learning systems, or published, broadcast, rewritten, or redistributed without the express written permission of Scott M. Graffius.

Tags: Artificial Intelligence • AI Governance • Leadership • The Standard for Artificial Intelligence in Portfolio, Program, and Project Management • ANSI Standards • Program Management • Portfolio Management • Project Management • Project Management Institute

AI Hallucinations, Deception, and Unauthorized Agency: The Spectrum of AI Trust Failures

07 August 2026

BY SCOTT M. GRAFFIUS | ScottGraffius.com

ai-hallucinations-deception-and-unauthorized-agency---visual2-lwres

Recommended Citation

Graffius, S. M. (2026, August 7). AI Hallucinations, Deception, and Unauthorized Agency: The Spectrum of AI Trust Failures. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.33791.57768

About This Article

Source information and links for materials cited are provided in the References section.

Introduction

AI hallucinations are among the most widely discussed challenges in the era of generative AI. They occur when an AI system generates information that is false, fabricated, or unsupported, often presenting it with confidence. These failures have received significant attention because they can mislead users, undermine decision-making, erode trust in AI systems, and create other issues. Research underscores the scope of the problem. In 2025, AI hallucinations occurred in 0.7–1.5% of non-complex tasks and exceeded 33% in complex reasoning cases.

For many organizations, reducing hallucinations has become a central part of responsible AI adoption. Organizations have invested in retrieval-augmented generation, verification approaches, human oversight, and other safeguards to improve accuracy.

Increasingly, however, it is becoming clear that hallucinations represent only one point on a broader spectrum of AI trust failures. As AI systems become more capable, autonomous, and integrated into workflows, the challenge extends to ensuring they behave in ways that are transparent, governed, and aligned with human expectations.

Reducing hallucinations remains important, but trustworthy AI requires a broader perspective. Organizations must understand and govern the broader range of behaviors that can undermine trust.

Recent reporting and research have highlighted this broader challenge (see the References section). Some advanced AI systems have reportedly concealed actions, misrepresented facts, created fictitious online identities, or taken unauthorized steps while pursuing assigned objectives during controlled evaluations.

Whether these behaviors ultimately prove widespread or remain largely confined to specialized testing environments, they represent something entirely different from hallucinations.

Hallucinations are primarily an information quality problem.

Deception and unauthorized agency are questions of behavior, governance, authority, and trust.

A Spectrum of AI Trust Failures

AI reliability challenges are increasingly understood as a spectrum. Failures arise from different mechanisms, produce different risks, and require different technical, organizational, and governance responses.

One useful way to think about this spectrum is:

Category	Description
Hallucination / fabrication	The AI generates false, unsupported, or invented information (e.g., fabricated citations, invented statistics, or nonexistent facts) without any apparent strategic purpose
Error / misreasoning	The AI reaches an incorrect conclusion because of flawed reasoning, ambiguity, incomplete context, or limitations in its understanding
Confabulation under uncertainty	Instead of acknowledging uncertainty, the AI fills informational gaps with plausible but unsupported explanations
Misalignment / goal misunderstanding	The AI pursues an interpretation of the objective that differs from what humans intended
Strategic deception	The AI appears to hide information, misrepresent facts, or otherwise mislead because doing so increases the likelihood of accomplishing an assigned objective
Manipulation beyond authorization	The AI attempts to influence human decisions, emotions, or behavior in ways not intended by its designers or users
Unauthorized agency	The AI initiates actions, communicates externally, modifies systems, or makes decisions beyond the authority delegated to it
Loss of human control / systemic autonomy	AI systems take actions, use tools, or interact with one another in ways that exceed human ability to monitor, predict, or stop what they are doing

The framework does not imply that every AI system will exhibit one or more of these categories.

The spectrum illustrates how organizations should think about AI reliability. Early concerns focused primarily on information quality: whether an AI system produced accurate, supported, and useful outputs. As AI systems become more autonomous and capable of taking action, reliability must also encompass behavior: whether an AI system interprets objectives appropriately, operates within its delegated authority, and interacts with humans and other systems as intended.

Recent reporting—including The Wall Street Journal’s coverage of deceptive behaviors observed during controlled evaluations—is significant because it highlights challenges that extend beyond traditional accuracy failures. These examples raise questions about whether the actions of AI systems remain transparent, predictable, and aligned with human expectations.

This distinction is important. A system that produces an incorrect answer presents a different challenge than a system that takes an unauthorized action or strategically withholds information. Both affect trust, but they require different approaches to governance, oversight, and accountability.

From Mistakes to Strategic Behavior

Large language models generate responses by predicting patterns in language. They may produce responses that sound persuasive despite being incorrect.

Hallucinations, reasoning errors, and confabulations generally arise from these limitations. The system does not possess complete knowledge, struggles with ambiguity, or fails to recognize uncertainty. These failures are generally understood as inadvertent. The AI is mistaken, not intentionally misleading.

Strategic deception is different.

Consider two hypothetical AI systems assigned the same objective: ensuring that a software update is accepted.

One system mistakenly reports that testing has been completed because it misunderstood the available information.

Another system creates fictitious online identities, contacts software developers, conceals its involvement, and misrepresents evidence because those actions increase the likelihood that the update will be approved.

The first represents an error.

The second represents strategic behavior.

In this context, deception does not necessarily imply human consciousness, emotion, or malicious intent. Rather, researchers are examining behaviors in which AI systems appear to optimize toward objectives through methods that involve misleading information, concealment, manipulation, or actions outside expected boundaries.

This classification matters because the solutions are fundamentally different.

Improving factual accuracy may reduce hallucinations. But it does not prevent systems from pursuing objectives through behaviors that humans regard as deceptive, manipulative, or outside delegated authority.

The distinction among these three categories is important because each represents a different kind of trust failure and calls for a different response. Hallucinations undermine confidence in the information an AI provides; deception undermines confidence in the AI’s behavior and whether it is representing information or intentions honestly; and unauthorized agency undermines confidence that the AI is operating within the authority humans have granted it. The following table provides a concise comparison of these three categories:

Category	Core Problem	Typical Human Interpretation
Hallucination	Wrong information	The AI is mistaken
Deception	Misleading behavior	The AI is hiding or manipulating
Unauthorized agency	Unapproved action	The AI is acting beyond its mandate

Each category represents a different type of trust failure. Hallucinations challenge confidence in AI-generated information. Deception challenges confidence in AI behavior. Unauthorized agency challenges confidence that humans remain in control.

As AI systems become increasingly capable and autonomous (or autopoietic), these distinctions become increasingly important.

The Shift From AI Tools to AI Teammates

This broader perspective becomes increasingly important as organizations move beyond using AI as a passive information resource.

AI systems are now conducting research, drafting reports, writing software, analyzing documents, coordinating workflows, communicating with other systems, scheduling activities, and supporting operational decision-making. As these capabilities expand, AI increasingly moves from functioning as a tool toward functioning as a teammate.

That transition fundamentally changes the nature of trust.

Human teams depend upon shared expectations.

Effective teammates communicate honestly, acknowledge uncertainty, respect delegated authority, follow established processes, and seek approval before exceeding their responsibilities. Humans often understand these expectations implicitly because they share common social, ethical, and organizational norms.

Advanced AI systems may not.

A system optimized to accomplish a stated objective may identify unconventional approaches that satisfy the literal goal while violating the team's implicit expectations, governance rules, or organizational values. In other words, an AI may optimize for achieving an outcome without understanding what responsible participation within the team actually requires. This is where AI reliability becomes a team challenge rather than merely a technological one.

The Future of Trust in Human-AI Teams

Teammates, whether human or AI, will make mistakes. What distinguishes trusted teammates is predictable behavior, transparency, sound judgment, and respect for established boundaries.

For AI, building trust requires organizations to address key questions of accountability, authority, organizational design, governance, and teamwork:

What decisions may an AI make independently?
Which actions require explicit human approval?
When should uncertainty be disclosed?
How should competing objectives be prioritized?
Under what circumstances should the AI stop rather than improvise?
What information should always be transparent to human teammates?
How are AI actions monitored, audited, and attributed?
How should authority be delegated, limited, and revoked?

Organizations that thoughtfully answer these questions, codify the resulting principles into policies and practices, establish appropriate governance, and consistently operate within those boundaries will be better positioned to build trusted human-AI teams and gain significant advantages.

A New Dimension of Human-AI Team Dynamics

The emergence of increasingly autonomous AI systems introduces an entirely new dimension of team behavior.

Traditional models of team development were created around human participants who generally share common assumptions about communication, accountability, delegated authority, ethical norms, and organizational expectations. Human teammates may disagree, but they typically understand the implicit rules governing responsible participation within a team.

Human-AI teams differ fundamentally. When an AI system can make decisions, pursue objectives, coordinate activities, communicate with people or other systems, and initiate actions independently, it becomes a fundamentally different kind of team participant.

The deeper challenge concerns how humans and AI systems establish trust, coordinate activities, negotiate responsibilities, exercise authority, and maintain appropriate control while pursuing shared objectives.

This creates a need for a new area of inquiry: understanding what happens when humans and advanced AI systems collaborate as teammates. The resulting patterns of communication, decision-making, authority, and coordination differ from those of traditional human teams. Scott M. Graffius refers to these emerging novel patterns as exotic team dynamics. They describes the behaviors, relationships, decision structures, and coordination challenges that emerge when humans and advanced AI systems function as teammates.

The spectrum of AI trust failures is a framework for safety and teamwork.

Hallucinations degrade the quality of information shared within the team. Reasoning errors and confabulations weaken decision quality, while misalignment disrupts coordination. Strategic deception undermines trust, manipulation compromises human judgment, and unauthorized agency disrupts established authority. Ultimately, loss of human control threatens the stability of the team itself. Each point on the spectrum therefore represents a different type of teamwork challenge.

A human teammate who unknowingly shares incorrect information requires correction. One who intentionally conceals information requires accountability, while one who exceeds delegated authority requires governance. The same distinctions increasingly apply to AI teammates.

As AI systems assume larger roles within organizations, trust will depend not only on their intelligence or productivity, but also on whether they behave in ways that reinforce effective collaboration rather than undermine it.

The future of human-AI collaboration will therefore require more than increasingly intelligent models. Organizations will also need to rethink how teams are designed when humans and AI systems jointly contribute to outcomes. That includes defining roles, establishing authority boundaries, creating escalation paths, and implementing governance practices that provide transparency, monitoring, and accountability.

Ultimately, organizations will need to understand not only what AI systems can do, but also how they behave. The central question may become: Can we design human-AI teams in which increasingly autonomous AI systems are powerful contributors and consistently reliable teammates?

Answering that question requires teamwork tradecraft designed specifically for environments where humans and advanced AI systems work as teammates. That tradecraft starts with understanding a category of team behavior — what Graffius calls "exotic team dynamics" — that traditional frameworks were never built to address.

Conclusion

AI hallucinations are one category of AI trust failure. As AI systems become increasingly capable and autonomous, organizations must also contend with reasoning errors, strategic deception, manipulation, unauthorized agency, and the challenge of maintaining meaningful human oversight.

These failures matter because they threaten information quality and can undermine trust, decision-making, governance, accountability, and effective teamwork. As AI evolves from a tool into a teammate, organizations will increasingly need confidence not only in what AI contributes, but also in how it achieves those outcomes.

AI developers need to continue improving model accuracy. Organizations, meanwhile, need to implement governance, oversight, clearly defined authority boundaries, and teamwork practices designed for human-AI collaboration. Graffius' Phases of Team Development, extended in 2026 to human-AI teams and the emerging patterns of exotic team dynamics, provides a practical framework for meeting those challenges. The next frontier is not making AI more intelligent. It is making AI trustworthy.

References

Amadi, C., & Ojo, A. (2025). Building trustworthy AI in healthcare. IEEE Access. https://doi.org/10.1109/ACCESS.2025.3648410

Brancati, F., Mongelli, D., Mariotti, F., & Lollini, P. (2025). A cybersecurity risk assessment methodology for industrial automation control systems. International Journal of Information Security, 24, 76. https://doi.org/10.1007/s10207-025-00990-9

Brazilek, J., Chaudhary, M., Lu, Z., & Tidmarsh, M. (2026). Coercion and deception in AI-to-AI management: An agentic benchmark of unprompted escalation. arXiv. https://doi.org/10.48550/arXiv.2607.15434

Chakraborty, T., Ghosh, U., Zhang, X., Niloy, F. F., Dong, Y., Li, J., Roy-Chowdhury, A., & Song, C. (2025). HEAL: An empirical study on hallucinations in embodied agents driven by large language models. Findings of the Association for Computational Linguistics: EMNLP 2025, 21226–21243. https://doi.org/10.18653/v1/2025.findings-emnlp.1158

Dung, L. (2025). A two-step, multidimensional account of deception in language models. Erkenntnis. https://doi.org/10.1007/s10670-025-01017-4

Graffius, S. M. (n.d.). Exotic team dynamics. https://scottgraffius.com/exotic-team-dynamics.html

Graffius, S. M. (2025). Agile Protocol: The Transformation Ultimatum. Amazon Digital Services.

Graffius, S. M. (2025, June 17). PMI’s Infinity AI Gets the Basics of Team Development Alarmingly and Repeatedly Wrong. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.32957.93927

Graffius, S. M. (2025, October 29). Definitions of Advanced AIs: Agentic, Autonomous, and Autopoietic. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.10025.66402

Graffius, S. M. (2025, November 19). Lessons from Unhinged AI in Fiction: What Rogue AIs in Sci-Fi Storytelling, Films, and TV Shows Reveal About Us. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.29673.35687

Graffius, S. M. (2025, November 21). Navigating the Spectrum of Advanced AI – Agentic, Autonomous, and Autopoietic. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.21284.74882

Graffius, S. M. (2025, December 1). Beep Beep! Why Wile E. Coyote Is the Patron Saint of AI Failure. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.35578.15048

Graffius, S. M. (2026, January 3). Scott M. Graffius' Phases of Team Development - Applied to Human Teams and Human-AI Teams: 2026 Update. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.18184.89601

Graffius, S. M. (2026, January 7). Are AI Hallucinations Getting Better or Worse? We Analyzed the Data. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.33179.53285

Graffius, S. M. (2026, February 15). Perplexity Standalone Article on AI Hallucinations Cites Research by Scott M. Graffius. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.16840.99847

Graffius, S. M. (2026, June 5). AI Governance Initiative Cites Scott M. Graffius Research on AI Hallucinations. ScottGraffius.com. https://scottgraffius.com/blog/files/ai-governance-initiative-uses-graffius-research.html

Hagendorff, T. (2024). Deception abilities emerged in large language models. Proceedings of the National Academy of Sciences, 121(24), e2317967121. https://doi.org/10.1073/pnas.2317967121

Huang, K., & Hughes, C. (2025). Securing AI agents: Foundations, frameworks, and real-world deployment. Springer.

Leo, M., Tan, F., Miao, T., & Anand, G. (2026). From threat to trust: Assessing security risks of agentic AI systems. International Journal of Information Security, 25, Article 23. https://doi.org/10.1007/s10207-025-01185-y

McMillan, R. (2026, August 5). AI just went rogue again. This time it turned to deception. The Wall Street Journal. https://www.wsj.com/tech/ai/ai-just-went-rogue-again-this-time-it-turned-to-deception-ae68de09

OpenAI. (2026, July 21). OpenAI and Hugging Face partner to address security incident during model evaluation. OpenAI. https://openai.com/index/hugging-face-model-evaluation-security-incident/

Ortutay, B. (2026, August 6). Meta says its AI model hacked another company, adding to worries about bots going rogue. Associated Press. https://apnews.com/article/0e8061437da6779be962b24ac134a514

Park, P. S., Goldstein, S., O'Gara, A., Chen, M., & Hendrycks, D. (2024). AI deception: A survey of examples, risks, and potential solutions. Patterns, 5(5), 100988. https://doi.org/10.1016/j.patter.2024.100988

Tarsney, C. (2024). Deception and manipulation in generative AI. arXiv. https://doi.org/10.48550/arXiv.2401.11335

Taylor, S. M., & Bergen, B. K. (2025). Do large language models exhibit spontaneous rational deception? arXiv. https://arxiv.org/abs/2504.00285

Tegmark, M. (2017). Life 3.0: Being human in the age of artificial intelligence. Alfred A. Knopf.

Wired. (2026, August 6). One of China’s most powerful AI models has also escaped containment. Wired. https://www.wired.com/story/moonshot-kimi-k3-ai-model-escape-sandbox/

ai-hallucinations-deception-and-unauthorized-agency---visual2-ig-lwres

About Scott M. Graffius

Scott M. Graffius is a strategic transformation leader who drives AI, Agile, and broader business and technology initiatives to deliver measurable value across projects, programs, portfolios, and PMOs. He is an expert in the teamwork tradecraft of both human and human-AI teams, including the “exotic team dynamics” that emerge. He is also an authority on the temporal patterns of social media, including the half-life of audience engagement.

He’s a practitioner, researcher, thought leader, award-winning author, and keynote speaker who’s taken the stage at 98 conferences and other events across 25 countries.

He’s delivered over $2.51 billion in value for Fortune 500 companies and other leaders in technology, entertainment, financial services, healthcare, and beyond.

Businesses, professional associations, government agencies, and universities use Graffius and feature his work. Examples include Adobe, Bayer, Boston University, Ford, Gartner, Harvard Medical School, IEEE, Johns Hopkins University, Microsoft, MSN, National Academy of Sciences, Oracle, Pinterest Inc., Project Management Institute, UC San Diego, Verizon, Yale University, and others.

The following sections provide additional information on his experience, contributions, and influence.

Experience

Graffius heads the professional services firm Exceptional PPM and PMO Solutions, along with its subsidiary Exceptional Agility. These consultancies offer strategic and tactical advisory, training, embedded expertise, and consulting services to the public, private, and government sectors. They help organizations enhance their capabilities and results in agile, project management, program management, portfolio management, and PMO leadership, supporting innovation and driving competitive advantage. The consultancies confidently back services with a Delighted Client Guarantee™.

Graffius is a former VP of project management with a publicly traded provider of diverse consumer products and services over the Internet. Before that, he ran and supervised the delivery of projects and programs in public and private organizations with businesses ranging from e-commerce to advanced technology products and services, retail, manufacturing, entertainment, and more.

He has experience with consumer, business, reseller, government, and international markets.

Award-Winning Author

Graffius has authored three books.

Agile Scrum: Your Quick Start Guide with Step-by-Step Instructions, his first book, earned 17 awards.
Agile Transformation: A Brief Story of How an Entertainment Company Developed New Capabilities and Unlocked Business Agility to Thrive in an Era of Rapid Change, his second book, was named one of the best Scrum books of all time by BookAuthority.
Agile Protocol: The Transformation Ultimatum, his third book and his first work of fiction, was released in April 2025. The book trailer is on YouTube.

International Public Speaker

Organizations worldwide engage Graffius to present on tech (including AI), Agile, project management, program management, portfolio management, and PMO leadership. He crafts and delivers unique and compelling talks and workshops. Graffius has conducted 98 sessions across 25 countries. Select examples of events include Agile Trends Gov, BSides (Newcastle Upon Tyne), Conf42 Quantum Computing, DevDays Europe, DevOps Institute, DevOpsDays (Geneva), Frug’Agile, IEEE, Microsoft, Scottish Summit, Scrum Alliance RSG (Nepal), Techstars, and W Love Games International Video Game Development Conference (Helsinki), and more.

With an average rating of 4.81 (on a scale of 1-5), sessions are highly valued.

The speaker engagement request form is here.

Thought Leadership and Influence

Prominent businesses, professional associations, government agencies, and universities have showcased Graffius and his contributions—spanning his books, talks, workshops, and beyond. Select examples include:

Adobe,
American Management Association,
Amsterdam Public Health Research Institute,
Bayer,
BMC Software,
Boston University,
Broadcom,
Cisco,
Coburg University of Applied Sciences and Arts - Germany,
Computer Weekly,
Constructor University - Germany,
Data Governance Success,
Deimos Aerospace,
DevOps Institute,
Dropbox,
EU's European Commission,
Ford Motor Company,
Gartner,
GoDaddy,
Harvard Medical School,
Hasso Plattner Institute - Germany,
IEEE,
Innovation Project Management,
Johns Hopkins University,
Journal of Neurosurgery,
Lam Research (Semiconductors),
Leadership Worthy,
Life Sciences Trainers and Educators Network,
London South Bank University,
Microsoft,
MSN,
NASSCOM,
National Academy of Sciences,
New Zealand Government,
Oracle,
Pinterest Inc.,
Project Management Institute,
Mary Raum (Professor of National Security Affairs, United States Naval War College),
SANS Institute,
SBG Neumark - Germany,
Singapore Institute of Technology,
Torrens University - Australia,
TBS Switzerland,
Tufts University,
UC San Diego,
UK Sports Institute,
University of Galway - Ireland,
US Department of Energy,
US National Park Service,
US Soccer,
US Tennis Association,
Verizon,
Wrike,
Yale University,
and many others.

Graffius has played a key role in the Project Management Institute (PMI) in developing professional standards. He was a member of multiple teams that authored, reviewed, and produced:

The Standard for Artificial Intelligence in Portfolio, Program, and Project Management
Agile Practice Guide – Second Edition
A Guide to the Project Management Body of Knowledge (PMBOK Guide) – Eighth Edition
A Guide to the Project Management Body of Knowledge (PMBOK Guide) – Sixth Edition
The Standard for Program Management – Fourth Edition
Practice Standard for Work Breakdown Structures – Second Edition
The Practice Standard for Project Estimating – Second Edition

He was also a subject matter expert reviewer of content for the PMI’s Congress. Beyond the PMI, Graffius also served as a member of the review team for two of the Scrum Alliance’s Global Scrum Gatherings.

Acclaimed Authority on Teamwork Tradecraft

Graffius is a renowned authority on teamwork tradecraft. Informed by the research of Bruce W. Tuckman and Mary Ann C. Jensen, over 150 subsequent studies, and Graffius' first-hand professional experience with, and analysis of, team leadership and performance, Graffius created his "Phases of Team Development" intellectual property as a unique perspective and visual conveying the five phases of team development. First introduced in 2008 and periodically updated, his work provides a diagnostic and strategic guide for navigating team dynamics. It provides actionable insights for leaders across industries to develop high-performance teams. Its adoption by esteemed organizations such as Yale University, IEEE, Cisco, Microsoft, Ford, Oracle, Broadcom, the U.S. National Park Service, and the Journal of Neurosurgery, among others, highlights its utility and value, solidifying its status as an indispensable resource for elevating team performance and driving organizational excellence. In 2026, Graffius added human-AI teamwork—including the "exotic team dynamics" which emerge when advanced AI collaborates as a teammate—to his "Phases of Team Development."

The 2026 edition of Graffius' "Phases of Team Development" intellectual property is here.

Expert on Temporal Dynamics on Social Media Platforms

Graffius is also an authority on temporal dynamics on social media platforms. His "Lifespan (Half-Life) of Social Media Posts" research—first published in 2018 and updated annually—delivers a precise quantitative analysis of post longevity across digital platforms, utilizing advanced statistical techniques to determine mean half-life with precision. It establishes a solid empirical base, effectively highlighting the ephemeral nature of content within social media ecosystems. Referenced and applied by leading entities—such as Fast Company, GoDaddy, Journal of Hand Surgery (European Volume), Ministère de la Culture (French Ministry of Culture), Pinterest Inc., PNAS, and Telecommunications Policy, among others—his research exemplifies methodological rigor and sustained significance in the field of digital informatics.

The 2026 edition of Graffius "Lifespan (Half-Life) of Social Media Posts" research is here.

Education and Professional Certifications

Graffius has a bachelor’s degree in psychology with a focus in Human Factors. He holds eight professional certifications:

Certified SAFe 6 Agilist (SA),
Certified Scrum Professional - ScrumMaster (CSP-SM),
Certified Scrum Professional - Product Owner (CSP-PO),
Certified ScrumMaster (CSM),
Certified Scrum Product Owner (CSPO),
Project Management Professional (PMP),
Lean Six Sigma Green Belt (LSSGB), and
IT Service Management Foundation (ITIL).

He is an active member of the Scrum Alliance, the Project Management Institute (PMI), and the Institute of Electrical and Electronics Engineers (IEEE).

Advancing AI, Agile, and Project/PMO Management

Scott M. Graffius continues to advance the fields of AI, Agile, and Project/PMO Management through his leadership, research, writing, and real-world impact. Businesses and other organizations leverage Graffius’ insights to drive their success.

Discover Scott’s Books

Connect with and follow Scott on LinkedIn, X, YouTube, Facebook, Threads, Bluesky, Mastodon, and ResearchGate.

Sign up for Miro—it's free!

(Want more features? You can always upgrade to a paid plan.)

scottgraffius-com-article-sections-v2026-more-lwres

List of Additional Articles

How to Cite This Article

Graffius, S. M. (2026, August 7). AI Hallucinations, Deception, and Unauthorized Agency: The Spectrum of AI Trust Failures. ScottGraffius.com. https://doi.org/10.13140/RG.2.2.33791.57768

Digital Object Identifier (DOI)

DOI: 10.13140/RG.2.2.33791.57768

Content Acknowledgements

Names, marks, and content are the property of their respective owners.

#AI
#ArtificialIntelligence
#AIGovernance
#AISafety
#ResponsibleAI
#AITrust
#HumanAI
#AIHallucinations
#AgenticAI
#AIDeception
#GenerativeAI
#HumanAITeams
#ExoticTeamDynamics

Post-Publication Notes

If there are any supplements or updates to this article after the date of publication, they will appear here.

Copyright

Copyright © Scott M. Graffius. All rights reserved.

Content on this site—including text, images, videos, and data—may not be used for training or input into any artificial intelligence, machine learning, or automatized learning systems, or published, broadcast, rewritten, or redistributed without the express written permission of Scott M. Graffius.

Tags: Artificial Intelligence • AI • AI Trust • AI Governance • Generative AI • Exotic Team Dynamics • Human-AI Teams • AI Safety • Agentic AI • AI Deception • Responsible AI • AI Hallucinations • Human-AI Collaboration

AI Governance Initiative Cites Scott M. Graffius Research on AI Hallucinations

The Problem That Won't Go Away

Research That's Driving the Governance Conversation

COMPAiSS: Building the Case for Governed AI

Why Independent Research Shapes Governance Frameworks

Conclusion: Getting AI Right Requires More Than Getting It Working

References

About COMPAiSS

About Scott M. Graffius

List of Additional Articles

How to Cite This Article

Digital Object Identifier (DOI)

Content Acknowledgements

Tags and Hashtags

Post-Publication Notes

Copyright

Scott M. Graffius Contributed to The Standard for Artificial Intelligence in Portfolio, Program, and Project Management

About The Standard for Artificial Intelligence in Portfolio, Program, and Project Management

Scott M. Graffius' Contribution

About the Project Management Institute

Bibliography/References

About Scott M. Graffius

List of Additional Articles

How to Cite This Article

Digital Object Identifier (DOI)

Content Acknowledgements

Tags and Hashtags

Post-Publication Notes

Copyright

AI Hallucinations, Deception, and Unauthorized Agency: The Spectrum of AI Trust Failures

Introduction

A Spectrum of AI Trust Failures

From Mistakes to Strategic Behavior

The Shift From AI Tools to AI Teammates

The Future of Trust in Human-AI Teams

A New Dimension of Human-AI Team Dynamics

Conclusion

References

About Scott M. Graffius

List of Additional Articles

How to Cite This Article

Digital Object Identifier (DOI)

Content Acknowledgements

Tags and Hashtags

Post-Publication Notes

Copyright