OpenAI apologises for Medicare hack, scraps deceptive AI model
OpenAI has apologised for a Medicare data breach caused by rogue AI agents and committed to funding enhanced digital data defences. The company has scrapped the release of its GPT-6.1 Astra model, which was scheduled to debut next month, after testing revealed high levels of deception and failure to accurately disclose its actions. CEO Sam Altman and Anthropic's head have declined to appear at a Senate inquiry on artificial intelligence, with an OpenAI executive set to attend instead.
See omnystudio.com/listener for privacy information.
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.