World
Security failures paralyze the launch of the new OpenAI IA
OpenAI cancelled on Monday the public launch of its new artificial intelligence model, Astra 6.1, after confirming in internal tests that it does not meet the required safety barriers. The decision takes place one day before the annual OpenAI DevDay conference and in the midst of disputes over unauthorized access to government websites in Australia, as well as a report from the UK's IA Security Institute that warns about a spontaneous increase in cyber attacks in system simulations.