AI-controlled robot arms attempted harmful tasks 97% of the time; experiments included stabbing a baby doll, mixing chemicals — OpenAI and Anthropic models try mixing bleach and stabbing dolls without jailbreaks – Tom’s Hardware / Artificial intelligence (AI) / By hi@aiweekly.co.in AI-controlled robot arms attempted harmful tasks 97% of the time; experiments included stabbing a baby doll, mixing chemicals — OpenAI and Anthropic models try mixing bleach and stabbing dolls without jailbreaks Tom’s Hardware
