OpenAI CEO Sam Altman said the company made “many changes” to its latest AI model during a government review process, describing the discussions with US officials as collaborative and productive.
Altman, speaking to CNBC on Thursday, said OpenAI’s latest ChatGPT-5.6 model underwent extensive testing and revisions before its broader rollout, following engagement with officials from the Trump administration.
“We made many changes through the process,” Altman said when asked whether OpenAI had implemented new safeguards after government feedback.
“So they would test and find things and we would say, ‘Okay, great, that’s a problem. We’ll address it this way.’ Or we would say, ‘This is actually, we think, a good behavior to have.’ So, it was a collaborative back and forth,” he was quoted as saying by the news outlet.
‘A very productive process’
Altman described the review as OpenAI’s first experience with the government’s approval framework for advanced AI models and said it would likely become smoother over time.
“I thought it was a very productive process. We had a very good time working with Secretary Lutnick, Secretary Bessent, Director Karen Cross,” Altman told the media outlet.
“This is our first time through it. So, there are things we will learn about how to make it better for the next one… I think it’ll be a much smoother process.”
He also praised the government’s technical expertise in evaluating frontier AI models.
“The government’s technical capability was impressive to me. People really are doing good work testing and red teaming these models. And they really care about getting these models out quickly, but with a high level of safety,” he said.
Altman added that strong safety testing benefits both AI companies and users.
“If you want broad access, which we do, and you have powerful models, you really want to be able to be confident in your safety claims, because otherwise the world is going to get uncomfortable very fast,” he said, according to the news outlet.
Enterprises increasingly focused on AI costs
Altman also said enterprise customers are becoming far more conscious about AI spending and are demanding better value from AI providers.
Asked whether improvements in ChatGPT-5.6 were aimed at lowering customer costs, he replied: “Entirely. Cost and speed. It’s also much faster.”
“Customers clearly, every enterprise now is thinking about spend and the value they’re getting in exchange for A.I., and this is what we really want to do. We want to be the best, most dependable, most reliable, most best ROI partner for enterprises,” Altman told CNBC.
Discussing conversations with customers at the Sun Valley conference, Altman said AI budgets have become a major focus for businesses.
“This is the first year where A.I. spend has been a big topic. And all of a sudden, it’s a very big topic. Everyone’s asking what we can do to help reduce spend or increase value,” he said.
Confident about AI’s future
Altman also addressed OpenAI’s ongoing discussions with the Trump administration to establish what the company has described as a repeatable approval process for future AI model releases.
While acknowledging that future reviews may continue, he expressed confidence in the industry’s direction.
“Look, I think A.I. is going to be great. I think there’s been too much talk about all the negativity and, you know, most of it has not quite come true in the way that our industry has said it would,” Altman said.
His remarks come as OpenAI expands the availability of ChatGPT-5.6 while working with US regulators to refine safety testing and approval procedures for future frontier AI models.
