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

One of the best tools we have to ensure that is that, like, if you have a company that hurts people, you’re liable for that.

Embedded evaluators are independent external experts placed within frontier AI companies to provide ongoing assessments and validation of a company’s safety, security, and operational systems, practices, and claims. [...] As frontier AI companies beg...

Australia should require AI companies to submit to independent safety checks.

In our national survey, 83% of U.S. adults expect AI to eliminate more jobs than it creates.

I don’t see current AI as dangerous. The calls to slow down AI development are a form of marketing and lobbying. Anthropic is portraying AI as highly dangerous to scare governments. But LLMs aren’t at that level yet.

That same kind of mechanistic understanding eludes us in today’s Super Intelligence systems, even as the frontier models powering these systems are now more capable than traditional software systems. We can’t attribute model behaviors and outputs to ...

I think we’ve got the beginnings of a pretty good story, because we can already build machines that can represent environments. [...] If that’s what you mean by consciousness—having beliefs, having desires, being able to report what those beliefs and...

I believe we were fired for prioritising safety over the near-term interests of OpenAI as a corporation.

At the information-processing level one can make the argument that LLMs do something very similar to what humans do: They extract patterns from language, compare them to other patterns of that language, and assign meaning.

Internal use: The same systemic-risk assessment applies to internal use of frontier models, including research, operational systems, and internal products. Internal deployment is assessed against the same capability thresholds and safeguards as pre-r...

AI is expected to automate a large number of jobs. But when large numbers of people lose their means of earning a living, the situation cannot stay unbalanced forever. [...] History has repeatedly shown that societies do not stay in extreme imbalance...

Allies should establish secure mechanisms to share information on serious AI incidents, emerging catastrophic capabilities, evaluation methods and effective safeguards.

Such a ruling close to Election Day would open a huge opportunity for abuse during a period where harm has outsized impact. Campaigns will use this technology to target voters in deceptive ways.

The first is a federal governance framework targeting frontier AI developers above $500 million in annual revenue. These developers would be required to publish safety frameworks, report critical incidents, and submit to semi-annual third-party audit...

An opt-out system as proposed by AI companies places burdens and costs on rights holders, as it would require ongoing surveillance to see whether copyright material is being used. We cannot scour the internet and ensure we are opting out on all of th...