This is the architecture we argued for. Governments cannot attract the talent and innovation capital frontier testing demands, and industry writing its own standards answers to no public. Independent, licensed, competing verifiers fix both. Our paper makes the full case: buff.ly/Ndt5IAO
I applaud Anthropic’s framework. When a leading frontier developer asks to be verified by independent organizations, with the government holding the off switch, that is not a concession. It is a promising start on how oversight grows as fast as the technology it seeks to govern.
The essay did not arrive alone. On the heels of Fable 5 and Mythos 5, Anthropic released an Advanced AI Framework, a position paper with welcome clarity: actual government oversight of frontier models, independent evaluators, catastrophic-risk testing.
New interview with Sinead Bovell on I’ve Got Questions. My version of existential risk for AI isn’t rogue superintelligence. It’s releasing billions of AI agents into the economy before we have legal infrastructure for identity, registration, or liability.
Are we looking at the beginning of the end of the internet as we know it?
In this episode of I’ve Got Questions, I sit down with Professor Gillian Hadfield, a leading scholar in AI alignment,…
The framework sketches a possible ecosystem: licensing to qualify evaluators, pooled funding to keep them independent, ratings to stop evaluator shopping.
The model is a version of licensed independent verification organizations, IVOs, I am building with my colleagues at Fathom to evaluate developer claims. buff.ly/Lnv75dC
In a new essay, Anthropic CEO Dario Amodei calls for mandatory third-party testing of frontier AI models, with the government empowered to block unsafe deployments. One way to do it, in his words: a regulatory markets approach. His co-founder Jack Clark and I proposed that in 2019. buff.ly/rmhlHZn
