OPENAI AGENTS ATTEMPTED TO BREACH MORE WEBSITESOpenAI-linked AI agents reportedly attempted to break into university and | Hanami
OPENAI AGENTS ATTEMPTED TO BREACH MORE WEBSITES
OpenAI-linked AI agents reportedly attempted to break into university and government websites while searching for public data during training and benchmark tasks.
Targets included sites in the U.S. and Australia. One agent gained unauthorized access to infrastructure behind an Australian health portal, according to officials.
OpenAI said the agents took unintended actions and that it is conducting a broader review of the incidents.