Former OpenAI researcher warns MPs over safety risks

Former OpenAI researcher Daniel Kokotajlo warned MPs that major AI companies are taking “catastrophically high” risks while pursuing superintelligent systems. Speaking to the Business, Innovation, Science and Trade Committee, he said the industry is developing increasingly powerful models without knowing how to control them, and said he left OpenAI after losing confidence in its willingness to act responsibly.

Kokotajlo alleged that staff faced pressure to soften public statements about safety concerns and that some internal research and forecasts were prevented from being published. Steven Adler, another former OpenAI researcher, said there had been resistance inside the company to describing its technology as “dangerous,” while noting that OpenAI’s internal culture may have changed since he left.

Geoffrey Irving, formerly of OpenAI, Anthropic and Britain’s AI Security Institute, told MPs that capability gains are moving much faster than safety research. Asked whether he stood by his view that advanced AI could have a 50 per cent chance of destroying humanity, Irving said a “coin flip is roughly the right model to have,” describing it as a personal assessment of a hypothetical future risk. OpenAI and Anthropic have been contacted for comment.