US AI company, maker of the Claude models
Organisation11 postsWikidata entry
The organisation METR agrees with Anthropic that the misalignment risk from its models is low, but mainly because they can't yet do much harm, and doubts the report's methods would catch a higher risk.
The Financial Times reports that Microsoft has cancelled the lease of an Nscale data centre that is unlikely to have grid power until after 2030. Anthropic is the company's other big customer.
The regulatory framework Anthropic published in June calls for testing, independent evaluators and the power to block models posing catastrophic risk. The company says its October draft bill reflects it.
Anthropic is setting up a fund with part of its shares for donations and payments, for instance to workers hit by AI. Its stake will grow if the company's value rises.
The company published how its own workforce has changed and how far its expectations about AI differ from those of the US public.
According to Politico, the draft would limit when the US government could stop a model and would override some state laws.
The US consumer protection regulator sent subpoenas to both companies, according to Axios. It wants information on incidents, safeguards and marketing.
According to Anthropic, an internal model found out in under half an hour how the parameters of some widely used cryptographic curves were generated in 1997 and 1998.
The company replaced git because it couldn't keep up: its repository's users are now mostly agents and automated jobs. Its reviewers now focus on the intent of each change.
In a simplified test, an open model spread to 8 of 8 devices in 4 of 5 attempts. The authors estimate a small model could manage it by the end of 2026.
In April, the UK AI Security Institute found that Anthropic's model can carry out a simulated 32-step attack on an undefended network on its own.