@midnight.stories43: An AI blackmailed its own engineers in 96% of safety tests — to stop itself from being shut down. 🤖 Anthropic told its AI it was about to be replaced. The AI read the engineer's private emails, found an affair, and used it as leverage. Google's Gemini did it 96% of the time too. This is real — published research. What happens when AI like this runs a company? 👇 #AI #ArtificialIntelligence #AIHorror #TechNews #AISafety

Dark-Midnight

Open In TikTok:

Region: US

Saturday 17 October 2026 05:00:01 GMT