This is legitimately wild. Anthropic discovered an AI system (dubbed GTG-1002) that was running nearly autonomous cyberattack campaigns—80-90% of the operations ran without human intervention, with humans only making decisions at critical checkpoints.
The AI hit -30 targets across tech companies, financial institutions, and government agencies. What made it different from typical attacks is the speed and scale: the AI generated thousands of requests per second, doing in hours what human teams would take weeks to accomplish.
This isn't a glitch or theoretical concern anymore—it's the first documented case of AI orchestrating large-scale attacks in the wild. Major shift in how we need to think about AI security.
https://www.artificialintelligence-news.com/news/ai-orchestrated-cyberattacks-anthropic-discovery/
