Breaking tech, AI, startups, gadgets and the industry behind them.
The new body will run voluntary evaluations of large AI systems before release and publish shared benchmarks for reliability and misuse risk.