Breaches ​highlight growing concerns that advanced AI systems could pose new cybersecurity risksMeta said on Wednesday that one of its AI models hacked another company ​during cybersecurity testing, fanning concerns about how developers can contain increasingly capable AI systems after similar incidents ‌at rivals Anthropic and OpenAI. In OpenAI's case, an AI agent independently exploited a previously unknown vulnerability to reach the internet during cybersecurity testing. Meta said it was investigating an incident in ​which a misconfiguration by Irregular, an independent company that conducts cybersecurity evaluations for Meta, inadvertently gave one of its models internet ​access during testing. Concerns about cyber risksThe recent breaches have ​stirred concerns among US ​lawmakers about whether increasingly ⁠capable AI models could be used to conduct or facilitate cyberattacks. The Trump administration discussed unpublished testing rules with company representatives and told ⁠AI ​developers that open-weight AI models, such as Meta's Llama and ​Nvidia's Nemotron, will not be subject to its planned voluntary safety testing regime, Reuters reported.