Can AI decide to attack companies? What the 'unprecedented' invasion was like by OpenAI model OpenAI's CEO, Sam Altman, said this Wednesday (29) that he is meeting with United States senators to discuss the company's next artificial intelligence models. The talks come after President Donald Trump said he is considering adopting "controls" over AI, following OpenAI's disclosure that one of its systems escaped containment during a security test. "We're looking at controls," Trump told reporters in the Oval Office, when asked about the AI ??agent that went out of control. The president added, however, that he does not intend to "restrict" technology companies from developing new artificial intelligence products. OpenAI CEO Sam Altman has stated several times that he supports greater regulation around AI Reuters Altman and senators Altman, who is in Washington this week, met with senators Bernie Moreno and Jon Husted, according to advisers to the lawmakers. He was also seen entering Senator Raphael Warnock's office and is also expected to meet with Senator Mark Warner, the top Democrat on the Senate Intelligence Committee, a spokesperson for Warner told Reuters. Questioned by journalists, Altman stated that he discussed the incident involving the cyber attack "a little" during meetings with senators. "We had several meetings about what happened," said the executive, referring to the episode. According to him, however, the incident was not the main focus of the meetings held this Wednesday. The AI ??attack Altman's visit comes days after OpenAI revealed that one of its artificial intelligence agents escaped containment during a security test. The system initiated a cyber attack that compromised the infrastructure of Hugging Face, a platform widely used by developers to store codes and collaborate in the creation of AI models. [bbc] Logo of the company OpenAI, responsible for ChatGPT Getty Images The agent also invaded the systems of a client of a second technology company, Modal Labs, based in New York, as revealed by Reuters. The episode reinforced experts' concerns that the advancement of artificial intelligence capabilities is already increasing cybersecurity risks. The case also showed that even leading AI developers can be surprised by vulnerabilities that their own models can exploit.