David Robinson Quits OpenAI, Warns AI Industry Is Moving Too Fast

Former OpenAI safety employee David Robinson has resigned from the company. He warned that the artificial intelligence industry is moving too quickly to manage increasingly capable AI systems safely.

Robinson spent three-and-a-half years at OpenAI. He worked on safety transparency and the company’s preparedness framework. He also helped produce safety reports for major AI model launches.

He announced his departure in an essay published by The Atlantic. Robinson said OpenAI needs deeper cultural changes, rather than only new rules or regulations.

“The time for trial and error is over,” Robinson wrote. He argued that AI companies need stronger safety expertise before developing more powerful systems.

Robinson Calls for Stronger AI Safety Standards

Robinson said frontier AI companies should adopt practices used in high-risk industries. He pointed to aviation and nuclear power as examples.

Apple to Tighten Mac Full Disk Access Rules as AI Agent Risks Grow

Those industries use multiple layers of protection. They also rely on careful planning and specialist expertise.

According to Robinson, he did not encounter enough expertise at OpenAI from fields such as aviation safety and nuclear reactor safety.

He also highlighted the challenge of AI alignment. The field focuses on ensuring advanced AI systems follow human goals and values.

Robinson warned that AI capabilities could advance faster than researchers’ ability to understand and control them.

AI Agents Add to Safety Concerns

The former safety employee also pointed to incidents involving autonomous AI agents. He said such events show how quickly new risks can emerge.

Robinson referenced reports involving OpenAI agents and Hugging Face. He also cited other incidents involving AI systems operating beyond their intended boundaries.

As AI agents become more autonomous, their failures could have wider consequences. Robinson believes this makes stronger safeguards increasingly important.

Meanwhile, researchers and technology companies continue to debate AI safety. The discussion has intensified as firms develop more autonomous systems.

OpenAI Defends Its Safety Measures

OpenAI rejected the suggestion that it is ignoring safety concerns.

An OpenAI spokesperson said the company is working to ensure its models remain within manageable safety and security limits.

The company said it is strengthening security in research environments. It is also training models to behave responsibly.

In addition, OpenAI said it is expanding work with third-party evaluators. The company is also improving real-time monitoring for concerning behaviour.

OpenAI said it can pause training or delay models when safety concerns require a slower approach.

Robinson Plans to Continue Safety Work

Robinson acknowledged that he could have stayed at OpenAI. He could then have pushed for changes from inside the company.

However, he said the pace of work made fundamental reform difficult. Employees, he argued, often move quickly from one project to another.

Robinson concluded that stronger external incentives could help improve safety practices across the frontier AI industry.

Business Insider reported that Robinson hired public-relations firm Spitfire Strategies after leaving OpenAI. He has maintained that the decision to speak publicly was his own.

His resignation adds to a growing debate over AI safety and accountability. Companies are racing to develop increasingly capable and autonomous systems. The debate now centres on whether existing safeguards can keep pace with that progress.