OpenAI CEO Sam Altman has defended his recent partnership with the Pentagon and addressed the concerns of staff at an all-hands meeting on Tuesday.
Recently, OpenAI has been under fire over its deal with the Pentagon, allowing the US military to use its AI tools for classified work. Prior to ChatGPT company, Anthropic was discussing the terms of a deal with the Department of Defense.
But, CEO Dario Amodei rejected the proposal over controversial demands, such as autonomous weapons and mass surveillance of American citizens, and OpenAI stepped in to make a deal.
As a result, the AI researchers and staff criticized OpenAI for yielding to the controversial demands of the Pentagon, thereby jeopardising the safe deployment of AI.
According to Altman, he stands by the deal with the Pentagon, but he regrets how quickly it was announced. As reported by the Wall Street Journal, he felt the timing made the company look “opportunistic” rather than staying consistent with the tech industry.
Altman said in the all-hands meeting, “To try so hard to do the right thing and get so absolutely like, personally crushed for it—and I know this is happening to all of you too, so I feel terrible for subjecting you all to this—is really painful.”
“I think this was an example of a complex but the right decision with extremely difficult brand consequences and very negative PR for us in the short term,” he added.
During the meeting, he also explained the government’s willingness to give power to OpenAI over the deployment of its technology and have a seat at the table for decision-making.
While defending the importance of a strong military for a country's national security, Altman said “The US military has been a great benefit to all of humanity over the last 250 years. Despite the great utility, he is not hesitant to show disagreement with their decisions.
OpenAI also disclosed that the company is looking to make a contract with NATO for classified networks. However, OpenAI later made a correction and said the opportunity would be for unclassified networks.