By Freddie Seba | AI Ethics & Governance for Leaders: Board & Trustee Guide to Responsible AI OversightResponsible AI Oversight
From GenAI to AI: Agency, Transparency, and the 2026 Board Agenda. © 2025 Freddie Seba
Year-end issue — closing 2025, setting the 2026 operating agenda
A quick year-end thank you (before anything else)
To everyone who took a risk this year—following, sharing, and thinking alongside the author while this newsletter was still “emerging”: thank you. You didn’t just subscribe to content. You joined a live social experiment in AI ethics and governance sensemaking, week by week, while the ground kept shifting.
The pivot (and why it’s not semantics)
In Issue #49 (2025 Year-in-Review), we made a deliberate pivot: GenAI is the loudest chapter—not the whole book.
Issue #50 completes that turn.
Because the governance question is no longer:
“What can it generate?”
It’s: “What can it do—at speed—under delegated authority—inside real institutions?”
And “real institutions” means leaders, boards, and trustees—always.
Even if this is an AI bubble… It’s still going to matter.
Here’s a core belief going into 2026: even if AI is a bubble—whether it pops, deflates, or just changes shape—AI will remain deeply relevant, because the underlying capabilities are already embedding into workflows, products, labor markets, education, healthcare, and institutional decision-making.
We’ve seen this movie before. The internet had hype cycles, crashes, and reinventions—then it became infrastructure.
So we move forward—not with hype—but with governance that holds across:
- GenAI (large language models and multimodal systems)
- Agentic AI (systems that can plan and execute tasks across tools/workflows)
- Embodied AI (systems operating in devices, robotics, autonomy)
- And the machine learning backbone is still driving “automation-by-default” incentives
The year-end thesis (building on Issues #48 + #49)
Put Issue #48 (partners, platforms, proof) together with Issue #49 (beyond GenAI), and the pattern is clear:
2026 is the year institutions move from advocating for AI to governing it—through rigorous evaluation, monitoring, and accountability.
Specifically:
- Agentic workflows move into core operations.
- Workforce disruption debates shift from “jobs” to human agency.
- Transparency declines right when oversight needs it most.
- Resilience becomes the control plane for systems that can act.
Who this issue is for (explicitly)
This newsletter is doubling down on trustees, boards, and executive leaders—because 2026 will reward institutions that treat AI as an operating condition:
Measurable. Auditable. Governed continuously. Owned at the top.
For presidents, provosts, deans, trustees, and executives navigating AI with public-trust accountability.
Board + trustee agenda for January 2026 (six items, no fluff)
If you put only six AI items on the first agenda of 2026, make them these:
- Define “proof.”
- Outcomes, safety, equity, learning quality—not just speed.
- Stand up an AI dashboard.
- Where AI is used, what it changes, incidents/near-misses, and concentration risk.
- Adopt a Human Agency standard.
- Task-level clarity on what must remain human-responsible—and where full automation is unacceptable.
- Make transparency contractual.
- Required disclosures, evaluation artifacts, monitoring, incident reporting—no “black box by default.”
- Treat AI agents as identity + operations risk.
- Visibility, policy constraints, approvals, monitoring, rollback/undo.
- Name dependency/lock-in risk.
- Second-source plans, exit clauses, portability requirements.
The Seba 12 Ps check (built to scale beyond GenAI)
- Purpose / Problems: What mission outcome are we improving—really?
- People / Protections: Who gets harmed when this fails—and how do we catch it early?
- Process / Policy: What operating controls govern day-to-day use (not just a PDF)?
- Privacy / Provenance: What data is in play—and where do outputs come from?
- Preparedness: Training, incident drills, escalation paths, and tabletop exercises.
- Product Ownership: Who owns outcomes, vendors, and lifecycle risk—end to end?
- Profits / Planet: Incentives and externalities (including scale effects leaders love to ignore).
Signals that will matter in 2026
1) AI economic dashboards (finally)
High-frequency “AI economic dashboards” track productivity, displacement, and new roles at the task level.
Board takeaway: This is governance instrumentation. If you can’t measure it, you can’t govern it.
2) Workforce shock is real—but governance is about agency
Workers tend to want AI for repetitive tasks while protecting judgment, oversight, and interpersonal work.
Emerging tools introduce a shared language for “green-light vs red-light” automation zones.
Board move: Stop debating “AI will change jobs.” Start making task-level decisions—automation vs augmentation—with explicit human involvement requirements.
3) Agents = the new resilience mandate
Agents are moving into end-to-end workflows; the governance gap is operational, not philosophical.
Board takeaway: Agent governance is a control plane—identity, logs, approvals, monitoring, rollback—not a policy memo.
4) Transparency is getting worse when we need it most
Corporate transparency about training data, compute, and post-deployment impacts is declining.
Board move: Transparency becomes a procurement gate for high-risk use cases—if the vendor can’t disclose, you can’t responsibly deploy.
Sector note (plain language): Healthcare evaluations are still too narrow
A key governance red flag: many studies evaluate LLMs as if they’re only taking tests.
But real healthcare deployment requires safety, bias/fairness, reliability, workflow fit, monitoring, accountability, and incident response.
Board lens: If evaluation is fragmented, governance is too.
What’s coming in 2026 (not just Issue #51)
If Issues #1–#50 were 2025’s knowledge build, 2026 is the conversion into board-ready operating practice—turning governance into muscle memory.
Across 2026 issues, expect reusable artifacts built for trustees, boards, and executive leaders:
- AI Oversight Charter (1 page, board-ready)
- AI Dashboard Template (monthly governance instrumentation)
- Human Agency Standard (task-level zones + approvals)
- Procurement + Contracting Addendum (disclosure, audit rights, monitoring, portability, exit clauses)
- Agent Governance Controls (identity, logging, rollback/undo, incident response)
If Issue #50 is the year-end signal, 2026 issues are the governance operating system.
Gratitude
Grateful to the institutions and communities that ground this work:
@University of San Francisco @StanfordHAI @Stanford Medicine @AMIA Informatics @Coalition for Health AI (CHAI) @UIC College of Applied Health Sciences @AAC&U
#AIGovernance #AIethics #ResponsibleAI #Boards #Trustees #HigherEd #Healthcare #DigitalHealth #RiskManagement #Compliance #Cybersecurity #Leadership
Transparency & copyright
Drafted and refined with generative tools for synthesis and clarity. Responsibility for interpretation, research, source selection, sensemaking, and conclusions remains with the author.
© 2025 Freddie Seba.
About the author
Freddie Seba is an author, speaker, and academic-practitioner with 20+ years across AI/GenAI, digital health, fintech, and higher education—focused on AI Ethics & Governance for Leaders and board-ready oversight.
MBA (Yale) • MA in International Policy Studies (Stanford) • EdD (USF)—degree conferral pending administrative processing following successful dissertation defense (Dec 10, 2025).*
The author brings board experience across B2B technology, digital health, and AI-enabled learning, with a practical emphasis on fiduciary oversight; AI risk, integrity, and privacy/security governance; and alignment across internal and external stakeholders—directly relevant to human- and society-centered leadership and responsible growth at scale.
Board and trustee briefings: connect on LinkedIn or visit freddieseba.com.
References and Useful Links
Stanford HAI — Stanford AI Experts Predict What Will Happen in 2026
https://hai.stanford.edu/news/stanford-ai-experts-predict-what-will-happen-in-2026
Google Cloud — AI agent trends 2026 report
https://cloud.google.com/resources/content/ai-agent-trends-2026
Google Cloud Blog — 5 ways AI agents will transform the way we work in 2026
https://blog.google/products/google-cloud/ai-business-trends-report-2026/
SiliconANGLE — AI agent governance is the new resilience mandate (Rubrik)
https://siliconangle.com/2025/12/23/ai-agent-governance-new-resilience-mandate-rubrikresilience/
Stanford HAI — Transparency in AI is on the Decline
https://hai.stanford.edu/news/transparency-in-ai-is-on-the-decline
Stanford HAI — What Workers Really Want from Artificial Intelligence
https://hai.stanford.edu/news/what-workers-really-want-from-artificial-intelligence
arXiv — Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce (WORKBank / Human Agency Scale)
https://arxiv.org/abs/2506.06576
JAMA — Testing and Evaluation of Health Care Applications of Large Language Models (systematic review)
https://jamanetwork.com/journals/jama/fullarticle/2825147
CHAI — Testing & Evaluation (T&E) Framework: EHR Information Retrieval
https://rai-content.chai.org/en/latest/electronic-health-record-information-retrieval/t%26e-framework.html
Tech Policy Press — Considering Nvidia’s Partnerships Push in Bid for Dominance
https://techpolicy.press/considering-nvidias-partnerships-push-in-bid-for-dominance