Daniel Kokotajlo told a UK parliamentary committee his group's proposal was the "least bad plan" on the market, while noting it still has problems, The Register reported on October 15. He is executive director of the AI Futures Project and a former OpenAI researcher who also testified at the New York City Council hearing earlier in October.
The hearing concerned whether the UK should add AI measures to its Cyber Security and Resilience Bill. The Register's article does not detail the plan he described.
Source: The Register