The Trump administration tried three times to block state AI laws. The FTC's July 1 policy statement was its latest attempt. Kevin Moriarty explains why it came up well short on preemption — and further than expected on everything else.

The FTC Statement on AI Bias Lacks Conviction

The FTC’s statement on AI bias stakes out an aggressive approach but fails to argue meaningfully for preemption of any state laws, writes Kevin Moriarty.

buff.ly

Also on BlueDot's Frontier AI Governance course, alongside the technical safety one. Technical alignment is getting one model to behave; governance is the harder problem: a world of them, built by rivals with no referee, not racing to the bottom. That second one is what I actually work on.

We've incorporated: Dissensus is now Dissensus Ltd (England & Wales, no. 17309927). A year of research and a stack of preprints now has a formal home to publish under and collaborate through. The work doesn't change, it just has a spine. → dissensus.ai

Starting BlueDot Impact's Technical AI Safety course this month. My research loop, if I'm honest: build an elaborate framework, test it in MARL, machine-check it in Lean, and realise around 2am I need to rewrite the whole paper. Time to be confidently wrong in front of people who catch it sooner.

Most composite crypto risk indices get sold as better alarms. Ours (with Andrew Maksakov) isn't. It barely edges the baselines, VIX still wins. The real value: a live, decomposable read on where stress comes from, DeFi vs TradFi. → https://arxiv.org/abs/2602.03874 #crypto #systemicrisk