OpenAI says it fired three safety researchers for mishandling sensitive information and breaking company rules. The researchers—Tomek Korbak, Jasmine Wang and Mikita Balesni—deny that account and say they were acting within their roles. The two sides disagree; the facts have not been independently established in the reports available.
The dispute centres on contact with outside safety experts, work on whether AI systems can be monitored, and Wang’s account that she accidentally opened a sensitive email after being given access for recruiting. OpenAI says an internal investigation found a wider pattern of misconduct. The researchers say the rules were unclear and that the way their dismissals were handled could make colleagues afraid to speak up.
So, is OpenAI’s safety culture already below an acceptable line? This dispute alone cannot prove that — but unclear rules, fear of retaliation and weak independent scrutiny would be warning signs. The researchers are speaking publicly, they say, because they fear silence will leave colleagues unsure whether raising concerns or consulting outside experts could cost them their jobs.
AI safety will depend not only on better technology, but also on people being able to question decisions, report risks and be heard by independent reviewers. If that fails, the risks could reach far beyond one company; if it works, scrutiny can help protect the public as AI grows more powerful.
WHEN SAFETY VOICES FALL SILENT, EVERYONE IS AT RISK.
Sanjay Sahay
Have a nice evening.