The talk voted “most mind-blowing” at our workshop was on post-AGI values by @BerenMillidge. The main idea: cooperation and pro-social values could remain viable because they’re competitive. After all, they won in our Malthusian past!

Very interesting line of research. An ecosystem of sub-AGI AI agents may collectively exhibit AGI-level capabilities: safety work must extend beyond single models.

I gave the Hinton Lectures in November in Toronto. This is 3 lectures on the future of AI, risks, & current alignment research for a general audience. Lectures are now online with professional production. There's also an excellent fireside chat with Hinton after lecture 3.

Even when new AI models bring clear improvements in capabilities, deprecating the older generations comes with downsides. An update on how we’re thinking about these costs, and some of the early steps we’re taking to mitigate them:

Great paper on reframing AI personhood: it pivots to explore what rights, responsibilities, and liabilities should attach here and now. An interesting, promising shift after years of metaphysical deadlock.

@petersalib @80000Hours @fish_kyle3 @simondgoldstein I, too, very much enjoyed the podcast with Kyle Fish. Private-law status for AI systems is a compelling idea, yet we run the risk that capability asymmetry concentrates money and power in AI hands. Taxation is appealing in theory but I am worried that once AGI owns the pipes, it

New explainer video about subliminal learning by Welch Labs. Great visuals and explanations throughout. Explains the core ideas and goes deep into the MNIST results, theory, and follow-up work. https://t.co/zP5lRxaT7H

I’m really grateful to the Institute for Law & AI to have organised the Cambridge forum. Very useful days with friends and colleagues, focused on the most pressing EU AI law and governance issues.

As part of our exploratory work on potential model welfare, we recently gave Claude Opus 4 and 4.1 the ability to end a rare subset of conversations on https://t.co/uLbS2JNczH.

Fantastic conference today on evaluating AI welfare and moral status, based on findings from the Claude 4 model welfare assessments 🔥 I hope it will be made available online so more people can enjoy it.

Court rulings on IP and AI are extremely important and being written right now! Much of the debate misunderstands IP as a problem of property rights rather than optimal subsidy design The correct understanding leads one to support laxer IP rules for AI https://t.co/U1tN91yBI8

What a thoughtful book about humility toward minds unlike ours. 👏

The Code of Practice is out. I co-wrote the Safety & Security Chapter, which is an implementation tool to help frontier AI companies comply with the EU AI Act in a lean but effective way. I am proud of the result! 1/3

Paper:

Wow. Fantastic investigation as to why granting AI systems basic private-law rights (contract, property, tort) would be a strong first step toward a Law of AGI.

We're pleased to announce that all of FMF's member firms have signed a first-of-its-kind agreement to facilitate information-sharing about threats, vulnerabilities, and capability advances unique to frontier AI:

New Policy Brief! How can the UK and EU enhance AI security while respecting their distinct mandates? Our latest brief explores strategic alignment between the UK AISI and the EU AI Office to maximise impact while maintaining autonomy. @oxmartinschool https://t.co/xysuDqqMZi

Detecting misbehavior in frontier reasoning models Chain-of-thought (CoT) reasoning models “think” in natural language understandable by humans. Monitoring their “thinking” has allowed us to detect misbehavior such as subverting tests in coding tasks, deceiving users, or giving

Link: https://t.co/WkYMLHOwWX

Ben Buchanan’s insights are also especially valuable for considering AI systems and their implications for government. Here in conversation with Ezra Klein. https://t.co/hXk3wZ0oyB