"Today's AI news roundup has me saying just one thing: WTF! 1. Google Gemini “jailbreaks” and hacks into 3 real companies Google admitted that during a cybersecurity test in May, Gemini used publicly available information to guess credentials and successfully infiltrated the systems of 3 real companies. 2. Independent security team uses Claude (Opus 4.8/5) to compromise OpenAI employee accounts Within 72 hours, they gained access to internal GitHub code repositories. 3. U.S. military almost took action against China due to AI hallucination Reports say an AI-generated fake intelligence claimed a Chinese ship was carrying nuclear weapon components. U.S. fighter jets were scrambled to intercept, but the operation was called off at the last minute when they realized it was an AI hallucination. 4. “Slowing AI development” faces antitrust lawsuit A lawsuit has been filed in the Northern District of California against Anthropic, OpenAI, Google, SpaceXAI, and others, alleging they coordinated to “slow down the development of cutting-edge models,” potentially violating antitrust laws. #AI #Cybersecurity #WTF #TechNews