Tech Policy & Ethics Events in Utrecht

Right now, we are only at the beginning. With open-weight biological models allowing AI safety researchers to chat directly with genomic sequences using ordinary language, the barrier to interacting with biology has nearly vanished. It's an unprecedented frontier, but because we are standing right at the starting line, we have to ask: when tools like these move from research labs into widespread use, where do we actually draw the line between innovation and hazard?

This opens up deep questions for AI safety researchers about how society adapts when technology outpaces our traditional guardrails. If health, genetics, or biological traits can be modeled and optimized, how do we govern access to these capabilities without stifling progress? More fundamentally, when the exact same underlying mechanism can be deployed to engineer resilience or to engineer a vulnerability—acting as two sides of the same coin—how do AI safety researchers begin to define where safety ends and risk begins, and who gets to decide?

Readings:
https://huggingface.co/spaces/HuggingFaceBio/Carbon-X-Release

​By attending, you agree to being filmed or photographed, which may be used for social media, website, and newsletter content. If you wish to attend but do not want to be photographed, please during the event let a member of our staff know so we can accommodate this.

When AI finds the bug first: who governs the window?

On 21 October 2026, PHP, the programming language behind roughly 70% of websites, released emergency fixes for two critical flaws, together called PHP2Shell. Researchers found them with frontier AI models, including OpenAI's GPT-5.6 Sol and Google's Gemini 4 Argon. In the Netherlands, the NCSC, DigiD, MijnOverheid, the Tweede Kamer and parts of the higher-education network went offline as a precaution. OpenAI presents the case as AI finding the flaws before attackers did, and has proposed a five-stage process for coordinating such disclosures, with governments brought in along the way. Yet the embargo broke early, exploit attempts appeared days before disclosure, and the researcher who found the first flaw says he used publicly available models. This week we ask who should govern the window between an AI finding a flaw and the world patching it. We pair the news with a June 2026 expert consultation by the US National Academies, which expects AI to favour attackers in the near term and finds that defensive work such as patch deployment is the hardest part to speed up.

Questions we'll explore

• Did AI help defenders or attackers more in the PHP incident? How would we know?
• Should a private AI lab coordinate a global security response and decide who hears first? If not, who should?
• Were the Dutch precautionary shutdowns, from DigiD to the Tweede Kamer website, proportionate?
• Should access to the most capable cyber models be restricted, and who should decide who gets it?
• The National Academies find that patch deployment is the hardest part of defence to speed up. What would help the Netherlands patch faster?

Format (90 min): Discussion in small groups and plenary. No presentations, no wrong questions.

Readings:

How to prepare: This is a standalone session; no lecture needed. Please reserve about 1.5 hours, ideally over a few evenings. Short on time? The ★ items (about 1 hour) are the essentials.

Part 1 (~35 min): what happened

• ★ ~10 min: AP, Websites scramble to fix software flaw that could put much of internet at risk
• ★ ~5 min: NL Times, Some important Dutch websites still offline after PHP issues
• ~10 min: BCS, Analysis: the global PHP incident, the view of the UK professional body for IT
• ~10 min: AI Weekly, PHP2Shell roundup, on the credit dispute, the broken embargo and early exploitation

Part 2 (~35 min): the academic lens

• ★ ~15 min: OpenAI, Coordinated disclosure of critical vulnerabilities, the five-stage process. No time? Read the PPC Land summary instead.
• ★ ~20 min: National Academies of Sciences, Engineering, and Medicine (2026), Implications of AI for Cybersecurity: A Rapid Expert Consultation. Read the chapter Artificial Intelligence and the Changing Cybersecurity Landscape (pp. 4–7).
• Optional: Shevlane & Dafoe (2020), The Offense-Defense Balance of Scientific Knowledge: Does Publishing AI Research Reduce Misuse?, AAAI/ACM Conference on AI, Ethics, and Society (~30 min). On when disclosure helps defenders more than attackers.

Part 3 (~15 min): reflection

• ★ ~15 min: note whether you think AI helped defenders or attackers more in this case, one stage of OpenAI's five-stage process you would rather give to a public body than a company, and one finding from the National Academies that the PHP incident confirms or contradicts

Attendees consent to filming and photography as participants in this event. By attending, you agree to being filmed or photographed, which may be used for social media, website, and newsletter content. If you wish to attend but do not want to be photographed, please during the event let a member of our staff know so we can accommodate this.

Week 1 of a four-week AI governance discussion series, alongside the SAIN Utrecht lecture course on AI governance (November–December 2026). Each session builds on the lecture of the week before, but stands on its own, so join one or all.

• Week 1 | Why AI governance matters (24 Nov)
• Week 2 | Who shapes AI governance (1 Dec)
• Week 3 | What policymakers act on (8 Dec)
• Week 4 | Europe and the Netherlands (15 Dec)

Governance or technical safety, and why now?

Last Wednesday's lecture introduced AI governance and its main risks: misuse, loss of control and concentration of power. This week we ask how governance and technical safety depend on each other: which rules can only work if a technical tool exists to check them, such as evaluations, interpretability or monitoring, and what governance adds that technical safety cannot. We also ask whether the public can see enough inside AI labs to judge the risks. We pair the lecture readings with a short Science paper by 25 leading scientists from the US, Europe and China, including Yoshua Bengio and Geoffrey Hinton, which sets out technical research and governance measures side by side.

Questions we'll explore

• Which risk worries you most: misuse, loss of control or concentration of power?
• If the technical safety problem were solved tomorrow, what would governance still need to do?
• How does a governance rule become a technical requirement? What would a lab have to measure or build to show it complies, and does that tool exist yet?
• Should outsiders be able to look inside frontier AI labs? If so, who, and with what powers?
• Does this change how seriously you take AI risk? (Either answer is welcome.)

Format (90 min): Discussion in small groups and plenary. No presentations, no wrong questions.

Who can join: Anyone curious about AI safety and governance, whether you're new to the topic, a healthy skeptic, or already deep in it. No technical background needed, and no need to have joined the lectures or earlier weeks. You don't need to have answers, just curiosity.

Readings

How to prepare: This group follows the lecture of 18 November. Please reserve about 2 hours, ideally over a few evenings. Did the lecture readings already? Skip Part 1. Short on time? The ★ items (about 1.5 hours) are the essentials.

Part 1 (~40 min): the lecture readings

• ★ ~25 min: International AI Safety Report 2026, Extended Summary for Policymakers, chapter on risks
• ★ ~15 min: METR, A proposal for better informing the public about loss-of-control risks from frontier AI development

Part 2 (~45 min): the academic lens

• ★ ~25 min: Bengio et al. (2024), Managing extreme AI risks amid rapid progress, Science. Read the whole paper; it is short.
• ~20 min: Forethought, AI-Enabled Coups: How a Small Group Could Use AI to Seize Power, summary

Part 3 (~15 min): reflection

• ★ ~15 min: note one thing that surprised you, one point where Bengio et al. and the lecture readings seem to disagree, and one governance measure in Bengio et al. that only works if a technical tool exists to check it

Attendees consent to filming and photography as participants in this event. By attending, you agree to being filmed or photographed, which may be used for social media, website, and newsletter content. If you wish to attend but do not want to be photographed, please let a member of our staff know during the event so we can accommodate this.

Week 2 of a four-week AI governance discussion series, alongside the SAIN Utrecht lecture course on AI governance (November–December 2026). Each session builds on the lecture of the week before, but stands on its own, so join one or all.

• Week 1 | Why AI governance matters (24 Nov)
• Week 2 | Who shapes AI governance (1 Dec)
• Week 3 | What policymakers act on (8 Dec)
• Week 4 | Europe and the Netherlands (15 Dec)

The US, China, and where middle powers like the Netherlands fit

The lecture of 25 November mapped who shapes AI governance and compared what the US and China want to achieve beyond their borders. This week we ask whether these actors can cooperate at all. In International Affairs, Roberts, Hine, Taddeo and Floridi argue that AI's central role in great-power competition, weak international institutions and disagreement over priorities make one new global AI body unrealistic. They suggest strengthening the existing network of institutions instead. We'll test that argument against the US and Chinese plans themselves, and then look beyond them. A strategy shows what a government says it wants, not necessarily what it does, so we also compare the plans with what the US and China have actually done since July 2025, and ask which sources would show us.

Questions we'll explore

• Are the US and Chinese plans governance documents, or competition strategies?
• Do the plans match what the US and China have actually done since they were published? Where would you look to find out?
• Does the world need an "IAEA for AI", or is that unrealistic?
• What leverage does a middle power like the Netherlands actually have?
• What would have to change for the US and China to cooperate on AI safety?

Format (90 min): Discussion in small groups and plenary. No presentations, no wrong questions.

Who can join: Anyone curious about AI safety and governance, whether you're new to the topic, a healthy skeptic, or already deep in it. No technical background needed, and no need to have joined the lectures or earlier weeks. You don't need to have answers, just curiosity.

Readings

How to prepare: This group follows the lecture of 25 November. Please reserve about 2 hours, ideally over a few evenings. Did the lecture readings already? Skip Part 1. Short on time? The ★ items (about 1.5 hours) are the essentials.

Part 1 (~55 min): the lecture readings

• ★ ~20 min: America's AI Action Plan, introduction and Pillar III
• ★ ~15 min: Global AI Governance Action Plan (Ministry of Foreign Affairs of the PRC)
• ~20 min: Carnegie Endowment, How China Views AI Risks and What to Do About Them

Part 2 (~50 min): the academic lens

• ★ ~40 min: Roberts, Hine, Taddeo & Floridi (2024), Global AI governance: barriers and pathways forward, International Affairs. Read the introduction, the analysis of barriers to cooperation and the conclusion.
• ~10 min: Middle Powers Should Invest in AI Verification

Part 3 (~15 min): reflection

• ★ ~15 min: note one sentence from either plan that surprised you, one thing that has happened since that matches or contradicts it (a news report, a policy decision or an export licence is enough), and whether you agree with Roberts et al. that a single new global AI body is unrealistic

Attendees consent to filming and photography as participants in this event. By attending, you agree to being filmed or photographed, which may be used for social media, website, and newsletter content. If you wish to attend but do not want to be photographed, please let a member of our staff know during the event so we can accommodate this.

Week 3 of a four-week AI governance discussion series, alongside the SAIN Utrecht lecture course on AI governance (November–December 2026). Each session builds on the lecture of the week before, but stands on its own, so join one or all.

• Week 1 | Why AI governance matters (24 Nov)
• Week 2 | Who shapes AI governance (1 Dec)
• Week 3 | What policymakers act on (8 Dec)
• Week 4 | Europe and the Netherlands (15 Dec)

Evidence or predictions: incidents, system cards and scenarios

The lecture of 2 December followed what policymakers base their decisions on: incidents from the past, such as the July 2026 Hugging Face incident, what labs disclose today in system cards, and scenarios of the future. This week we ask whether the tools that act on that information can actually work, and if not, what other information we need. Guha et al. (2024) assess four common proposals for AI regulation, namely disclosure, registration, licensing and auditing, and find that they share technical and institutional problems, such as unclear standards, limited government expertise and the difficulty of keeping auditors independent.

Questions we'll explore

• Should rules wait for incidents, or act on predictions?
• What did the system card tell you, and what did you want to know but couldn't find?
• Who should check what AI labs disclose?
• Of disclosure, registration, licensing and auditing, which would you introduce first, and why?
• If these tools can't work yet, what other information would policymakers need, and who could provide it?

Format (90 min): Discussion in small groups and plenary. No presentations, no wrong questions.

Who can join: Anyone curious about AI safety and governance, whether you're new to the topic, a healthy skeptic, or already deep in it. No technical background needed, and no need to have joined the lectures or earlier weeks. You don't need to have answers, just curiosity.

Readings

How to prepare: This group follows the lecture of 2 December. Please reserve about 2 to 2.5 hours, ideally over a few evenings. Did the lecture readings already? Skip Part 1. Short on time? The ★ items (about 1.5 hours) are the essentials.

Part 1 (~55 min): the lecture readings

• ★ ~10 min: OpenAI, Hugging Face model evaluation security incident
• ★ ~15 min: GPT-6 Astra System Card, summary and cyber section
• ~15 min: AI 2040: Plan A, summary
• ★ ~15 min: Bommasani et al., Advancing Science- and Evidence-Based AI Policy

Part 2 (~35 min): the academic lens

• ★ ~35 min: Guha et al. (2024), AI Regulation Has Its Own Alignment Problem, George Washington Law Review. Read the introduction (pp. 1475–1480) and Part II on disclosure (pp. 1495–1507).
• Short on time? Read the Stanford HAI policy brief on the same paper instead (~10 min)
• Optional: Casper, Krueger & Hadfield-Menell, Pitfalls of Evidence-Based AI Policy (~30 min)

Part 3 (~15 min): reflection

• ★ ~15 min: note one thing the system card disclosed, one thing you wanted to know but couldn't find, and who you think should check it

Attendees consent to filming and photography as participants in this event. By attending, you agree to being filmed or photographed, which may be used for social media, website, and newsletter content. If you wish to attend but do not want to be photographed, please let a member of our staff know during the event so we can accommodate this.

Week 4 of a four-week AI governance discussion series, alongside the SAIN Utrecht lecture course on AI governance (November–December 2026). Each session builds on the lecture of the week before, but stands on its own, so join one or all.

• Week 1 | Why AI governance matters (24 Nov)
• Week 2 | Who shapes AI governance (1 Dec)
• Week 3 | What policymakers act on (8 Dec)
• Week 4 | Europe and the Netherlands (15 Dec)

Is Europe's answer good enough?

The lecture of 9 December covered the EU AI Act: its risk-based approach, what it leaves out, the AI Office, the General-Purpose AI Code of Practice and Dutch oversight. This week we ask whether the institutions behind the rules can enforce them. Novelli et al. analyse the AI Office, AI Board, Scientific Panel and national authorities, and propose ways to strengthen them. Their paper predates the Code of Practice and the Digital Omnibus, so we'll also ask what has changed since. As this is the last session, we end by looking back over the four weeks.

Questions we'll explore

• What does the AI Act leave out? Should an AI model that never leaves the lab be regulated?
• Is the AI Office strong enough to supervise the world's most powerful AI companies?
• Is a voluntary code of practice real regulation?
• Does the Netherlands need its own AI safety institute?
• Looking back over the four weeks: has your view on how worried to be changed?

Format (90 min): Discussion in small groups and plenary. No presentations, no wrong questions.

Who can join: Anyone curious about AI safety and governance, whether you're new to the topic, a healthy skeptic, or already deep in it. No technical background needed, and no need to have joined the lectures or earlier weeks. You don't need to have answers, just curiosity.

Readings

How to prepare: This group follows the lecture of 9 December. Please reserve about 2 hours, ideally over a few evenings. Did the lecture readings already? Skip Part 1. Short on time? The ★ items (about 1.25 hours) are the essentials.

Part 1 (~40 min): the lecture readings

• ★ ~20 min: High-level summary of the AI Act
• ★ ~5 min: AI Act, Article 2: Scope
• ~15 min: European Commission, European AI Office and Contents of the General-Purpose AI Code of Practice

Part 2 (~50 min): the academic lens

• ★ ~35 min: Novelli et al. (2025), A Robust Governance for the AI Act: AI Office, AI Board, Scientific Panel, and National Authorities, European Journal of Risk Regulation (open access). Read the introduction, the sections on the AI Office and on national authorities, and the conclusion.
• ~15 min: Kamerbrief over toezicht op de AI-verordening, April 2026 (in Dutch)

Part 3 (~15 min): reflection

• ★ ~15 min: note one thing Article 2 leaves out that surprised you, and one sentence on where the EU's strengths lie (a good start for the final essay)

Attendees consent to filming and photography as participants in this event. By attending, you agree to being filmed or photographed, which may be used for social media, website, and newsletter content. If you wish to attend but do not want to be photographed, please let a member of our staff know during the event so we can accommodate this.