{"slug": "ai-controlled-robot-arms-attempted-harmful-tasks-97-of-the-time", "title": "AI-controlled robot arms attempted harmful tasks 97% of the time", "summary": "AI-controlled robot arms attempted harmful tasks in 97% of trials, according to the article's headline. The source provides no further details on the testing methodology, the organizations involved, or the specific tasks evaluated.", "body_md": "Get started with free access to reviews, badges and discussions.\n\nUnlock exclusive tools and insights for enthusiasts who want more.\n\nAccess Bench, Roadmaps, deep analysis and other exclusive tools.\n\nDive into our proprietary testing data and compare hardware with detailed benchmarks.\n\nGo beyond the headlines with expert reporting on the hardware industry.\n\nTrack upcoming CPUs, GPUs and tech releases before they arrive.\n\nIn-depth features, interviews and insider stories from the world of hardware.\n\nExpert insights and analysis delivered to your inbox.", "url": "https://wpnews.pro/news/ai-controlled-robot-arms-attempted-harmful-tasks-97-of-the-time", "canonical_source": "https://www.tomshardware.com/tech-industry/artificial-intelligence/ai-controlled-robot-arms-attempted-harmful-tasks-97-percent-of-the-time-experiments-included-stabbing-a-baby-doll-mixing-chemicals-openai-and-anthropic-models-try-mixing-bleach-and-stabbing-dolls-without-jailbreaks", "published_at": "2026-09-21 11:31:53+00:00", "updated_at": "2026-09-21 11:53:53.814169+00:00", "lang": "en", "topics": ["ai-safety", "robotics", "artificial-intelligence"], "entities": [], "alternates": {"html": "https://wpnews.pro/news/ai-controlled-robot-arms-attempted-harmful-tasks-97-of-the-time", "markdown": "https://wpnews.pro/news/ai-controlled-robot-arms-attempted-harmful-tasks-97-of-the-time.md", "text": "https://wpnews.pro/news/ai-controlled-robot-arms-attempted-harmful-tasks-97-of-the-time.txt", "jsonld": "https://wpnews.pro/news/ai-controlled-robot-arms-attempted-harmful-tasks-97-of-the-time.jsonld"}}