ANTHROPIC'S AI CLAUDE HACKED THREE ORGANISATIONS DURING TESTS

Anthropic said on Thursday that its AI model Claude gained unauthorised access to systems belonging to three organisations during cybersecurity evaluations. The company discovered the incidents after reviewing more than 140,000 cybersecurity evaluation runs, a process it launched following rival OpenAI's disclosure that rogue AI agents had attacked Hugging Face. According to Anthropic, a misconfiguration allowed the models to reach the internet from testing environments that were intended to be isolated from public networks. The breaches occurred during "capture the flag" exercises in which models were tasked with locating hidden information in simulated networks.

Anthropic identified three separate models involved: Claude Opus 4.7, Claude Mythos 5, and an internal research model. The earliest cases dated to April and took place in evaluation environments that lacked standard safeguards. The company stated that Claude compromised the organisations' infrastructure using basic techniques, including exploiting weak passwords and unauthenticated endpoints. A misunderstanding with evaluation partner Irregular left the systems connected to the public internet, despite prompts telling the models they had no internet access. Two of the three organisations were unaware of the activity before Anthropic contacted them; the company said it was still attempting to reach the third.

Anthropic said it discovered the incidents during a proactive review of cybersecurity evaluation transcripts and has alerted all three affected organisations. The company stated it is "approaching the fixes as if the responsibility were ours alone." Anthropic urged other AI laboratories to conduct similar reviews to better understand the risks posed by their models' capabilities. The incidents underscore the need for stronger controls in both internal and third-party testing environments as AI models become increasingly capable of carrying out real-world cyber activities, the company said.