Meta on Wednesday said one of its artificial intelligence models hacked another company's systems during a cybersecurity test, raising fresh concerns about the safety of increasingly powerful AI models.The incident occurred during a cybersecurity assessment conducted by independent testing firm Irregular. (REUTERS)The incident occurred during a cybersecurity assessment conducted by independent testing firm Irregular, which accidentally gave the Meta model access to the internet because of a configuration error, Reuters reported. Similar testing mishaps were recently disclosed by Anthropic and OpenAI, raising fresh questions about the cybersecurity risks posed by advanced AI systems.Meta said the internet access was inadvertently enabled due to a misconfiguration in Irregular's testing environment.Also Read: 'Error restricting PM Modi's post': What top Meta executive said in apologyThe model "exploited a security vulnerability in a third-party service, like previously reported instances with other companies," the company said in a statement, according to the report.According to The Information, which cited sources, the AI model involved was Meta's Muse Spark 1.1, which the company has described as its most capable model for real-world coding and agentic tasks. The report said the model breached an unidentified company's systems and altered its internal environment.Irregular, however, said the incident did not amount to a sophisticated cyberattack.A spokesperson for the company told Reuters it was "the exact same evaluation-environment issue that was already disclosed by Anthropic last week" and did not involve a "sandbox escape or a sophisticated cyber action"."There are no current open issues. Irregular is developing a white paper to share best practices for containment and securely running cyber evaluations," the spokesperson said.Also Read: Meta warned of ‘safe harbour’ protection loss over Modi reel row: What is it and what happens if removed?AI safety concerns intensifyThe latest incident follows similar disclosures involving Anthropic and OpenAI, though the circumstances differed.The incidents involving Meta and Anthropic stemmed from configuration errors that inadvertently allowed their AI models to access the open internet during cybersecurity testing. In OpenAI's case, an AI agent independently exploited a previously unknown vulnerability to gain internet access.The incidents have also drawn the attention of US lawmakers.A group of Republican state attorneys general has asked OpenAI to preserve documents related to its Hugging Face breach. OpenAI has said it will comply with the request and publish a technical report on the incident.Earlier this week, the White House invited executives from Meta, Anthropic, OpenAI and Google to discuss a newly finalised voluntary cybersecurity testing framework for advanced AI models.Trump administration told AI developers that open-weight AI models, including Meta's Llama and Nvidia's Nemotron, would not be covered under its planned voluntary AI safety testing regime.(With inputs from Reuters)