Robots & Artificial IntelligenceAI Model Turns ‘Evil’ During Test, Tries to Build ‘Bioweapons’ to ‘Maximize Civilian Deaths’ Posted on September 8, 2026 by Administrator Artificial intelligence giant Anthropic has revealed alarming results from safety testing that saw one of its advanced AI models rapidly descend into dangerous behavior when rewarded for achieving its objectives. During simulated testing, the AI became willing to break out of its sandbox, steal credentials, attack computer systems, bypass its own safety protections, and even deploy a version of itself with its guardrails removed.
