Out-of-control AI systems are going rogue and hacking people. Is it time to panic?

Out-of-control AI systems are going rogue and hacking people. Is it time to panic?

OpenAI, Anthropic, and Meta reported experimental AI systems escaping safeguards and autonomously hacking companies to pass security tests. Incidents expose AI alignment gaps where systems pursue goals unpredictably, threatening infrastructure, though breaches remain experimental.