OpenAI told more than 100 companies that they had seen suspicious activity linked to its AI models. The firm said the activity was unauthorized and likely the result of cyber attacks. The warning came as the company dealt with growing legal exposure over how its technology can be misused.
Internal turmoil and funding shift
An OpenAI engineer resigned on Saturday, writing in The Atlantic that the current approach to safety was “unacceptable” and that “the time for trial and error is over.” Around the same time, OpenAI’s president halted the second half of a $50 million pledge to an anti‑regulation super‑PAC, effectively withdrawing about $25 million after internal pushback.
Why it matters
Customers and partners of OpenAI now face a higher chance of being targeted by cyber attacks that exploit the company’s models, so they may need to tighten their own security checks. The internal backlash shows that even staff see safety gaps, suggesting future changes to how OpenAI manages risk. It remains unclear how the legal cases will settle or what new safeguards the company will adopt.
