The closed-door briefing came just days after the company disclosed a cyber breach that exposed internal communications, adding urgency to the discussions.
The new model, which Altman described as a significant advancement over the company’s existing systems, was shown to a bipartisan group of senators and House members. Attendees said the demonstration highlighted the model’s ability to reason through complex problems and generate more nuanced responses than previous iterations, though specific technical details were not shared publicly.
Altman’s visit to Washington underscored the growing tension between the rapid pace of AI development and the government’s ability to regulate it. Lawmakers have been pressing for clearer rules around the technology, particularly after the breach raised concerns about how OpenAI protects its proprietary systems from malicious actors.
The cybersecurity incident, which the company acknowledged in a recent blog post, involved an attacker gaining access to internal messaging systems. OpenAI said the intruder did not access the core AI models or customer data, but the episode has fueled calls for stronger safeguards as the company prepares to deploy more powerful tools.
During the briefing, Altman emphasized that safety testing and external audits would be central to the new model’s rollout. He also reiterated the company’s support for federal legislation that would require transparency around AI capabilities and risks, a position that aligns with some of the bills being drafted in both chambers.
Several lawmakers who attended the session said they were impressed by the model’s performance but remained cautious about its potential for misuse. “The technology is moving faster than our oversight,” one senator told reporters afterward, speaking on condition of anonymity to discuss the private meeting.
The demonstration comes as OpenAI faces increasing scrutiny from regulators in the United States and Europe. The European Union is finalizing its AI Act, which would impose strict requirements on high-risk systems, while the Federal Trade Commission has opened an inquiry into the company’s data practices.
Altman left the Capitol without taking questions from the press. A spokesperson for OpenAI said the company would provide more details about the new model in the coming weeks, including a timeline for public release and additional security measures implemented in response to the breach.
