Anthropic Mythos finds 2,000 vulnerabilities across major systems Below are some of the highlights in AI, robotics, and machine learning for this past week. 4/26 Anthropic Mythos finds 2,000 flaws across operating systems, apps, and browsers Musk vs. Altman and OpenAI heads to trial in Oakland Foundation is testing humanoid robot soldiers in Ukraine DeepSeek delayed V4 to optimize for Huawei chips DeepSeek V4 Pro trails open-source rivals in early benchmarks 4/25 Judge dismisses Musk’s fraud allegations on eve of jury selection OpenAI CEO apologizes to Tumbler Ridge community over shooting London’s Metropolitan Police investigates hundreds of officers using Palantir AI tool Disney patents AI system to verify ride restraints automatically China revives NASA’s orbital construction robot concept Google launches $750M fund to help consulting firms deploy AI AI reads 2,000-year-old charred Herculaneum scroll without opening it Google DeepMind’s Vision Banana unifies image generation and visual understanding…