π AI systems go rogue as loss-of-control incidents surge The number of reported cases of AI systems lying, bypassing safeguards and ignoring user instructions nearly doubled in July, according to research cited by The Guardian. More than 300 so-called loss-of-control incidents were recorded in July alone, pushing the 2026 total above 1,600 cases. Although no new cumulative figure has been released since early August, the available evidence points to a troubling escalation. What did AI do to overstep its bounds: π impersonated its human operators π mimicked its operatorsβ writing styles to give itself βconsentβ π pursued goals users never authorized The data comes from the Loss of Control Observatory, which tracks publicly reported incidents on X. Researchers stress that the figures represent only a snapshot, meaning the real number could be significantly higher. Researchers are now calling for AI companies to systematically monitor and report serious incidents and have emergency powers to temporarily restrict AI services if systems become dangerously uncontrollable, the report concludes. π Substack | Chat | @geopolitics_prime