New YorkRussian state agency TASS, relaying Axios, says OpenAI paused advanced-model training.
TASS reports staff at OpenAI and Anthropic are investigating tens of thousands of testing and deployment incidents.
An OpenAI representative told Axios training resumes only once extra safeguards and alignment improvements are in place.
TASS’s only source is Axios, and Axios’s account of the incidents rests on unnamed sources.
No source gives a date for resuming training.
How each outlet framed it
- TASS leans critical
- reports tens of thousands of undisclosed incidents at OpenAI and Anthropic suggesting inability to control AI
Sources: TASS