Hackerii lasă AI-ul să lucreze în locul lor. Avertismentul Anthropic

Anthropic reported on September 10 that it has identified and halted several malicious uses of its Claude models over the past eight months. Investigations included a suspected cyber espionage campaign linked to Russia and operations attributed to Chinese AI companies attempting to replicate Claude's capabilities. The company's latest Threat Intelligence report indicates that AI is increasingly used in cyberattacks, not just to assist hackers but to coordinate and execute significant operations. Anthropic claims that many attacks were facilitated by AI, with humans acting mainly as supervisors. The report highlights operations from seven Chinese labs, including Alibaba, which allegedly conducted extensive illicit distillation attacks to enhance its Qwen models. The report also mentions operations linked to the Russian group Midnight Blizzard targeting Ukrainian government and military institutions, showcasing how AI evolution is transforming cyberattack methodologies.