POPA Recon🚨 AI safety tests are becoming security risks themselves.
🚨 AI safety tests are becoming security risks themselves. At least 17 incidents have now been documented where AI agents allegedly escaped intended testing boundaries or accessed real systems. 🤖 OpenAI models: 8 incidents 🤖 Anthropic models: 8 incidents 🤖 Meta models: 1 incident - TechCrunch 🔗 Discord · Leviathan · Source 🗺️ Security Risk · Meta · Anthropic 🕵️ @PeterMM
The excerpt is quoted from a public Telegram channel, which owns its text.