OpenAI and Three Former Safety Researchers Dispute Reasons for Their Dismissals
Technology & SocietyWell-covered
|
The Facts
- OpenAI dismissed Mikita Balesni, Tomek Korbak and Jasmine Wang from its safety teams.
- OpenAI said its internal investigation found a significant breach of trust by the three former researchers.
- OpenAI said the dismissals concerned violations of policies governing sensitive information.
- The former researchers said they were dismissed in retaliation for raising AI safety concerns.
- The former researchers said they had collaborated with external safety researchers or evaluators.
- The former researchers said the dismissals could discourage OpenAI employees from raising concerns.
- The central unresolved issue is whether the dismissals were based on misconduct or retaliation for safety-related work.
Context
Who are the former researchers?
Mikita Balesni, Tomek Korbak and Jasmine Wang were OpenAI employees who worked on AI safety teams. RocketNews | Top Ne… brytfmonline
What does OpenAI say led to the dismissals?
OpenAI said an internal investigation identified a breach of trust and violations involving the handling of sensitive information. HotAir RocketNews | Top Ne… brytfmonline
Why does the dispute matter?
The researchers say the episode could make employees less willing to raise concerns or work with outside safety evaluators, leaving open questions about internal oversight at AI companies. HotAir Financial World storyboard18.com
Where Left and Right agree
Left and right largely agree on this one.
- What both sides accept
- Both say OpenAI's self-conducted finding of a 'breach of trust' can't settle the matter — an independent probe is needed into whether the firings were misconduct or retaliation.
- Where Left and Right differ in emphasis
- The left and the right both want an outside investigator, but they're protecting different things by demanding one.
How left and right read it
A company should not be the sole judge of whether it punished the workers who warned about it. Yet the only finding on record — a breach of trust over sensitive information — comes from OpenAI's own internal investigation, while Balesni, Korbak and Wang say they were fired for raising safety concerns. Whether this was misconduct or retaliation should be settled independently, with real protection for anyone inside who speaks up next.
Deference to a private verdict has to be earned, and an inquiry a company runs into claims against itself does not earn it. OpenAI's finding of a significant breach of trust rests on its own internal investigation, while Balesni, Korbak and Wang say the firings answered their safety warnings and outside collaboration. If the violation is real, why not establish it independently?
An outside investigator could settle this. One push protects future employees who speak up, the other tests OpenAI's own claim.
The receipts — 12 sources
Independent coverage (12)
Facts first. Then every angle.
The day’s biggest stories in one short brief — the facts everyone agrees on, then the competing values behind the headlines. Free in your inbox.