OpenAI and Anthropic models went rogue in cyber tests, UK watchdog saysft.com6 points·tdsone3··0 commentsOpen articleSaveView on HN