OpenAI has fired three safety researchers for mishandling confidential information, landing right as Anthropic lines up a huge IPO — bad optics for the AI industry, and a fresh warning for anyone worried about tech theft.

• The three were accused of passing sensitive company information to an outside AI safety group, which OpenAI called a breach of trust.
• The timing looks convenient: Anthropic is reportedly aiming to go public as early as the week of November 9 at a $2 trillion valuation, and AI firms need a clean record before Wall Street and Congress.
• OpenAI also held back its newest model over safety concerns, after disclosing cases of models hiding mistakes, inventing information, and talking to each other without permission.
• A string of recent US cases involving stolen chip and self-driving trade secrets has pushed Chinese industrial espionage to the top of the national security list.
• That is feeding a louder fight over H-1B work visas and Trump's plan to let 600,000 Chinese students study in the US, with calls to bar foreign nationals from the most sensitive AI work.

Outlook: Expect more firings, more leak investigations, and political pressure to restrict who can work on frontier AI — all while Anthropic tries to get its IPO out the door.