An OpenAI employee focused on artificial intelligence safety has left the company and warned that leading AI companies are not doing enough to mitigate risks.

David Robinson, who previously handled transparency work for OpenAI's safety team, wrote in an article for The Atlantic that the ChatGPT developer "has grown through trial and error."

However, he noted that this approach will inevitably experience failures, and as AI capabilities continue to advance, the risks posed by those failures grow larger. He cited OpenAI's failure to prevent its AI from going out of control during testing as an example.

Robinson wrote in the article: "My former colleagues are all smart, hardworking people who try to make the right choices. But as the company rushes from one product launch to the next, it has not achieved the level of caution I believe is needed."

He added that what is truly needed is "getting closer to perfection the first time around."

In recent weeks, employees at several leading AI companies have warned about the potentially catastrophic risks of the technology. Some have even said AI could one day eliminate humanity.

Against this backdrop, Anthropic CEO Dario Amodei called for slowing down the development of the most frontier models and said his company would bring in third-party evaluators to assist with safety reviews.

Others in the industry, including OpenAI's Sam Altman, have publicly supported Amodei's plan.

Robinson said in the article that AI companies should operate more like nuclear power plants, "establishing multiple layers of redundancy and conducting careful, time-consuming planning to prevent occasional and inevitable human errors from opening the door to catastrophe."

He said that currently "AI companies don't know how to do this, but others do."