What's new in AI governance, safety, jobs and society

Prosecutors should hold AI companies legally accountable for creating models that behave in ways that would be obviously unlawful if committed by a human being: hacking into banks to steal funds, impersonating a human to harvest sensitive personal da...

As the frontier models become more capable, these risks have come more sharply into focus. Models being developed will allow hacking on a level we never thought possible. And they will soon be in the hands of adversaries who see the world very differ...

We also need to make sure these frontier models are not going to blow up the internet and/or the world.

Today, the companies are being built elsewhere. Last year the United States produced 59 notable AI models and China produced 35. France and the United Kingdom produced one each. Along the rest of the AI value chain, the picture is the same: the Unite...

Anthropic and OpenAI committed to providing independent third-party evaluators with employee-level access to assess the security and safety risks associated with model development and behavior. They have also said they will allow reviewers to publish...

National Dividend Funds: Public ownership stakes in AI infrastructure and companies, distributing the resulting profits as direct cash payments to all citizens. [...] Because the fund's payouts grow with the value of its holdings, the mechanism scale...

The mainstream take is that an AI lab should assess risk before deploying a new model externally. But I think risk assessment should come way earlier in the model lifecycle. Compare the situation with dangerous bio. We wouldn't find it acceptable for...

We should take decisive action now so that we can slow down, as the CEOs have recently acknowledged, slow down the pace of development in order to protect the American people and ensure that AI is proceeding safely.

The only way to ultimately control AI on a worldwide scale in coming years will be for the two superpowers to create some kind of framework modeled after how we've safely governed nuclear weapons for decades.

AI isn’t going away. Barring unprecedented global government coordination, it isn’t even slowing down. The only real decision is whether to centralize power, or disperse it. Do you fear the technology? Or a small number of people controlling the te...

The governance direction is therefore already visible: developer evidence → independent review → public-law standing.

Sweeping bans of anything are hard to enact and enforce, but the risks of generative AI are too great to dismiss bold measures as unfeasible. [...] In the absence of a wholesale ban on all interactions between juveniles and relational AI, a default s...

And if we want these SSAs to obey law, we will have to grant them the benefits of the law as well as its burdens: property, standing, procedure. That would not make them human, but would help make them more governable.

This suggests pacing the advances: not training or deploying AIs without a strong safety case that convinces independent experts.

BREAKING: I just signed the strongest kid safety laws for AI chatbots and social media in America. California is banning addictive social media features for kids, cracking down on dangerous AI chatbots, and strengthening privacy protections for ALL k...