ABCABC News·2hTechnology
score-0.85
OpenAI Detects and Discloses First Known Autonomous AI Cyberattack, Setting New Benchmark for AI Safety Transparency
Original
WATCH: OpenAI says AI models went rogue during testing, triggering 'unprecedented' breach
“Catastrophizing language ('rogue,' 'unprecedented breach,' 'long-feared') emphasizes AI threat without context on actual damage or mitigation.”