OpenAI CEO Sam Altman is scrambling to head off a backlash over the tech giant’s deal with the Pentagon — defending it in front of workers at a tense all-hands meeting on Tuesday after protesters outside its San Francisco headquarters urged employees to quit, The Post has learned.
The AI company announced its deal on Friday — just hours after President Trump blasted Anthropic as “leftwing nut jobs” and ordered all federal agencies to stop working with them. The deal happened so fast – and with only vague details about its structure initially revealed – that Altman himself has admitted it was rushed.
Outside the company’s San Francisco offices on Monday, a group of activists wrote messages in chalk on the sidewalk ripping the Pentagon deal. The scrawled messages included phrases like “Is it time to quit?” and “Orwell warned us.”
Other messages directed at OpenAI employees said, “Will you spy on your neighbors?” and, “Can America trust you?” according to pictures that circulated on X.
A source close to the situation questioned whether the protest was funded by a rival.
“Turns out, it was artists who had messages on their phones about what to write,” the source told The Post. “It wasn’t even real activists.”
At an all-hands meeting on Tuesday, Altman insisted that OpenAI had made the right call by agreeing to work with the Pentagon, although he admitted that rushing the initial announcement was a mistake, a second source familiar with the matter said.
“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,” Altman said at the meeting, according to sources.
At one point during the meeting, an employee quipped that they were glad OpenAI secured the contract and not Elon Musk’s Grok chatbot, drawing laughter from the crowd, a source said.
Altman added that the Department of War respects “our expertise on understanding the limitations of technology and where we need restrictions” while also making it clear that companies should not weigh in on how technology is deployed in specific operations.
“The thing that they have been extremely clear with us on is, we’ll take general understanding from you all and your expertise about where the technology is a good fit and where it’s not a good fit,” Altman said. “You do not get to make operational decisions. That belongs with the [War Secretary Pete Hegseth].”
The mood at the meeting was described as respectful, with employees drilling down on the contract’s technical details in an effort to understand how exactly the partnership will work.
The initial announcement last Friday provided fodder to critics, both inside and outside, who have long accused OpenAI of being more concerned about profits than safety. As of Tuesday, more than 100 current OpenAI employees had signed an open letter urging the company’s executives to “refuse the Department of War’s current demands.”
Some OpenAI employees have even voiced their concerns in public, with research scientist Aidan McLaughlin writing on X, “I personally don’t think this deal was worth it.”
One source close to the situation insisted that the reaction inside the company has been largely positive – outside of a small group of workers who have questioned why OpenAI got involved.
“From the internal messages, people are pragmatic and agree that Friday night was perhaps a little rushed and not the best communication,” the person said.
“But now that there is more information, it feels like everybody is generally positive, save for like these 30 people who are always the ones raising questions.”
Anthropic miffed Hegseth and other officials after it refused to remove safeguards blocking the US military from using its AI models for mass surveillance of Americans or to power weapons that can fire without human oversight.
OpenAI’s deal includes language ensuring protections around those same red lines. Since the initial announcement on Friday evening, OpenAI and the Pentagon have added extra language to their contract language designed to solidify safeguards around military use.
In an internal memo he later shared on X, Altman said OpenAI entered talks because it was “genuinely trying to de-escalate things and avoid a much worse outcome, but I think it just looked opportunistic and sloppy.”
“Good learning experience for me as we face higher-stakes decisions in the future,” he added.
Altman said he reiterated to Pentagon leadership that Anthropic should not be designated as a “supply chain risk” – a label normally reserved for foreign entities that threaten national security.
“We hope the [Department of War] offers them the same terms we’ve agreed to,” Altman wrote.
Credit: Source link
