Ex-OpenAI security employee accuses company of negligence

A leading security employee at AI company OpenAI has accused his former employer of negligence after being fired.

Companies like OpenAI are “not remotely careful enough,” wrote ex-employee David Robinson in the US magazine The Atlantic. He called for significantly greater security standards to be established for the development of artificial intelligence (AI).

To date, OpenAI has developed its products based on the principle of “trial and error,” Robinson wrote. Errors were therefore only corrected if they occurred after a new language model was published. This time is now over: AI systems are now so powerful that similarly high security standards must apply to them as for other high-risk technologies. “Advanced AI research must be run like a nuclear power plant or a busy airport,” Robinson wrote.