Discussion Group - week 4 - What's being done about AI Safety

Part four of a four-week AI safety discussion series: Week 1 | Why AI Safety matters · Week 2 | What failure could look like at scale · Week 3 | Why alignment is hard · Week 4 | What's being done about it? Every session stands on its own, so join one or all.

Who it's for: 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 earlier weeks. You don't need to have answers, just curiosity.

Discussion group curriculum:
https://docs.google.com/document/d/160PpLjFN9SaZja0njeKIQHnD-M1Uy4718dNRDyhaucs/edit?usp=sharing

After looking at why AI Safety matters, how failure could look at scale, and why alignment is hard, this week we ask a hopeful and honest question: what is actually being done, and is it enough?

We'll compare three kinds of response. Governments could regulate frontier AI developers. Developers could make structured arguments that their systems are safe to deploy. And countries with very different values could try to agree on shared standards.

Questions we'll explore

• Which approach do you find most convincing, and why?
• Who should decide when an AI system is safe enough to release?
• These papers are a few years old: what has changed since, and do their ideas still fit?

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

Readings:
★ = mandatory (~2 hours). The rest is optional.

• Frontier AI Regulation: Managing Emerging Risks to Public Safety

• ★ Executive summary, the "building blocks" of regulation, and the proposed initial safety standards

• Optional: the rest of the paper

• Safety Cases: How to Justify the Safety of Advanced AI Systems

• ★ Abstract, introduction, and the overview of the four types of safety argument

• Optional: the worked examples of each argument type

• Confucius, Cyberpunk and Mr. Science: Comparing AI ethics between China and the EU

• ★ The whole paper, skipping the appendix and references

• International AI Safety Report 2026: https://internationalaisafetyreport.org/publication/international-ai-safety-report-2026

• Optional: §3.2 Risk management practices

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.