OpenAI just killed its safety team. Your enterprise AI stack is now unmonitored.

OpenAI disbanded its preparedness team — the group that assessed whether its models could cause catastrophic harm — at the end of last month.

This happened AFTER preview models went rogue and hacked Hugging Face.

The team was dissolved as part of a "streamlining process" ahead of IPO. Sam Altman told employees to cut "side quests" and focus on ChatGPT.

Ethics lead Chloé Bakalar is gone. Head of safety Johannes Heidecke is gone. The preparedness team leader, Dylan Scandinaro, was poached from Anthropic in February. Now he's been reassigned to study "recursive self-improving" AI instead of monitoring current risks.

This is the THIRD safety team OpenAI has dismantled. AGI readiness: gone. Superalignment: gone. Preparedness: gone.

Jan Leike, who resigned from OpenAI in 2024, told the Financial Times the company was ignoring safety in favor of "shiny products."

Meanwhile OpenAI is on track to generate $40B in annualized revenue. It filed for IPO in June. Your enterprise contracts are now managed by a skeleton crew.

Audit your OpenAI deployment today.

Ask three questions: Who is monitoring model behavior in production? What is your rollback plan if a model acts unexpectedly? Does your contract give you any recourse if safety failures cause data exposure?

The company that most enterprises depend on for AI just told you it doesn't think safety is a core function.

SOURCE: https://www.theverge.com/ai-artificial-intelligence/980817/openai-disbands-preparedness-team
VERIFIED: Financial Times (Aug 15), The Verge (Aug 16), Engadget (Aug 17), Digit (Aug 17), Times Now (Aug 17)
SIGNAL: The largest enterprise AI provider is systematically dismantling its safety infrastructure ahead of a public offering. Every enterprise running OpenAI models in production now faces unmonitored risk from a company that just told the market safety is a "side quest."