← Back to live feed1 story
Fired OpenAI researchers Jasmine Wang, Tomek Korbak, and Mikita Balesni urged the company board to halt development of models with hard-to-monitor reasoning in a letter sent on Wednesday. The former safety researchers argued that OpenAI and industry rivals should stop pursuing such models, claiming their dismissals on Oct. 1 created a chilling atmosphere for remaining employees. OpenAI terminated the three employees after an internal investigation found they mishandled sensitive data, a decision the company says was unrelated to safety concerns.
Sign in to suggest edits
Key sources
- SOURCE@zeffmax“three recently fired employees say the company and its competitors should not move forward with work that would decrease AI monitorability”x.com
- SUPPORT@keachhagey“Korbak served as the technical contact for Redwood Research and METR in their investigation of the Hugging Face incident”x.com
- SUPPORT@hesamation“the firings are “chilling everyone still at OpenAI””x.com
- SUPPORT@thezvi“If you fire safety researchers for sharing confidential information with a safety org, and you move so quickly you don't have a statement”x.com
- SUPPORT@jachiam0“a very difficult moment for trust in the AI labs to do the right thing where it comes to independent oversight and verification”x.com
- SUPPORT@fleetingbits“if this is true though, this was essentially whistleblowing; openai basically covered up these incidents”x.com
- SUPPORT@_nathancalvin“Tomek and Mikita were the lead authors of the extremely influential Chain of Thought Monitorability paper”x.com
- SUPPORT@mark_k“Working on 'AI safety' doesn't give you a free pass to leak confidential information”x.com