Days after triumphantly declaring victory and striking a big deal with the U.S. government, Sam Altman is stuck in his defense.
In an interesting development, OpenAI’s recent fiasco is perhaps the least technically challenging yet. This also means that Altman can’t just push app updates to alleviate the issue.
Altman and OpenAI decided last week to enter into a deal with the Department of Defense for access to AI models after rival Anthropic refused to agree to the government’s terms, and are now in damage control mode.
A fast-paced tale of contract negotiations, optics, and the tipping of the scales in a chatbot war, the public watched everything unfold by downloading and voting with their wallets.
So how did we get here? Whether you’re looking to catch up or are simply confused as to where the drama started in the area, we’ve got you covered.
This is last week’s OpenAI Rocky live commentary.
Tuesday, February 24th: Anthropic is given an ultimatum.
Anthropic CEO and Co-Founder Dario Amodei Chance Yeh/Getty Images for HubSpot
Before this story was about OpenAI, it was about Anthropic CEO Dario Amodei, Altman’s former colleague and current rival. (If you want to know how intense their feud is, the two refused to hold hands during a recent photo shoot with other AI executives).
For several days, the Pentagon continued to threaten Antropic, accusing it of denying the military unfettered access to its AI systems. As tensions reached a fever pitch, Defense Secretary Pete Hegseth summoned Amodei to an in-person meeting at the Pentagon.
A Pentagon official later told Business Insider that Hegseth gave Amodei an ultimatum. If they didn’t board the ship by 5:01 p.m. ET on February 27, they would essentially be blacklisted by the U.S. government.
Wednesday, February 25: OpenAI begins its own consultations
A person familiar with the situation told Business Insider that OpenAI began discussions with the Department of Defense on Wednesday.
But for now, all eyes are on whether Anthropic will comply with the Pentagon’s demands or risk losing business and potentially facing other repercussions.
Thursday, February 26: Anthropic CEO details red lines for AI companies to cross in meeting with Department of Defense
Amodei is a biophysicist by training and is known for passionately pursuing the safe development of AI. In fact, that’s why he and the other six co-founders, including his sister Daniela, say they left OpenAI to start their own company in the first place.
Amodei said Anthropic wanted to work with the U.S. military, but needed safeguards to ensure its technology was used safely in two specific cases.
human
Amodei said Anthropic “cannot knowingly provide” AI used to guide fully autonomous weapons, and said such technology could not be reliably deployed in combat. Amodei was also concerned that Anthropic’s AI could be effectively used to conduct large-scale domestic surveillance, even if such searches did not violate the letter of the law.
“We cannot in good conscience comply with their request,” Amodei said in a statement. He expressed hope that the Pentagon would reconsider its threat about what would happen to Anthropic if it did not acquiesce.
In response, Emile Michael said, “@DarioAmodei is a liar and has a God complex,” the head of the Pentagon’s research and development department wrote to X.
that’s a shame @DarioAmodei He is a liar and has a god complex. He wants nothing more than to try to personally control the U.S. military, and he is comfortable putting our nation’s security at risk.
of @DeptofWar We always abide by the law, but we don’t bow to the whims of commercial technology… https://t.co/ZfwXG36Wvl
— Undersecretary of the Army Emile Michael (@USWREMichael) February 27, 2026
Friday, February 27: President Trump slams Amodei and humanity
Amodei did not have to wait long for a response after rejecting the reported final Pentagon proposal. Just an hour before the deadline, President Donald Trump took down Anthropic, calling the AI startup a “radical leftist, crazy company.”
“I am directing all federal agencies of the United States government to immediately cease using Anthropic’s technology,” President Trump told Truth Social. “We don’t need it, we don’t want it, and we won’t do business with them again!”
Hegseth then said he intended to formally declare Anthropic a supply chain risk to national security, a designation that would effectively bar companies with Pentagon contracts from doing business with the AI startup. (Antropic responded that such a blanket designation is “unprecedented” and that it intends to challenge it in court if it is enacted.)
Altman signs contract Anthropic doesn’t want
Late Friday night, Altman announced on X that OpenAI had signed a contract with the Department of Defense to “deploy our models to sensitive networks.”
Tonight, we reached an agreement with the Department of the Army to deploy our model on a classified network.
In all of our interactions, DoW demonstrated a deep respect for safety and a desire to partner to achieve the best possible outcomes.
AI safety and widespread distribution…
— Sam Altman (@sama) February 28, 2026
He also said he does not support the idea that the U.S. government would designate Anthropic as a supply chain risk and has made his position clear to officials.
Although some OpenAI employees publicly celebrated the deal, backlash quickly ensued within the AI community, including from former top OpenAI employees.
Miles Brundage, former head of policy at OpenAI, said of “By making it seem like it wasn’t an expedition, they were ruining Anthropic while helping Anthropic.”
Government officials’ claims that OpenAI would allow the Pentagon to use its AI models for “all lawful” purposes quickly caught people’s attention, with some criticizing it for leaving the door open to use cases that Anthropic said were red lines it didn’t want to cross.
For the avoidance of doubt, I would like to add that OpenAI – @DeptofWar The agreement flows from the “all lawful uses” touchstone that the DoW rightly advocated and xAI agreed to. But as Sam explained, it refers to certain existing legal authorities and includes certain mutually agreed upon… https://t.co/boVlh5kgfc
— Senior Secretary Jeremy Lewin (@UnderSecretaryF) February 28, 2026
Saturday, February 28th: ChatGPT defectors help push Claude to #1 on the App Store
As OpenAI endured criticism over the deal, Anthropic’s popularity soared. Some ChatGPT users decided to vote with their mobile phones and took refuge with Anthropic’s Claude chatbot.
Expressions of support for Claude quickly led to downloads.
On Saturday, Claude became the top-downloaded app on Apple’s U.S. App Store. A few weeks ago, Claude wasn’t even in the top 40, but Anthropic beat out competitors like OpenAI’s ChatGPT, Google’s Gemini, and xAI’s Grok.
Pop star Katy Perry endorsed Anthropic and shared a screenshot of her signing up for the company’s $214.99 annual Claude Pro plan.
Altman goes on defense.
Amid mounting criticism, Altman answered questions from users about X in a private AMA Saturday night, acknowledging that the deal with OpenAI was “definitely rushed and the optics aren’t great.”
Ultimately, Altman said he hopes the agreement with OpenAI will “de-escalate the situation.” Only time will tell whether it’s the right move, he said.
The optics are not in good condition as we were obviously in a hurry.
We wanted to calm things down and thought the deal we were offered was a good one.
If we’re right and this leads to a détente between Dow and the industry, we’ll look like geniuses and become the next company…
— Sam Altman (@sama) March 1, 2026
Monday, March 2: Altman says OpenAI has amended the government agreement
Following President Donald Trump’s statement, the Treasury Department announced it would no longer use Anthropic’s products.
in the direction of @POTUS, @USTreasury will terminate all use of Anthropic products, including use of the Claude platform within the department.
The American people deserve to have confidence under President Trump that all tools of government serve the public good… https://t.co/R7rF0ci5CY
— Treasury Secretary Scott Bessent (@SecScottBessent) March 2, 2026
Altman later announced that OpenAI had amended the contract to add more specific protections to prevent the Pentagon from using OpenAI’s models to conduct large-scale domestic surveillance.
“Protecting the civil liberties of Americans is important and we were very focused on this, so we wanted to make this point especially clear by including commercially obtained information. As with everything we do in iterative deployments, we will continue to improve as we learn,” he wrote in an internal message later published by X.
Importantly, this language still cites current U.S. law, but Amodei reiterated that the law is outdated and, even if it does not violate the law, does not adequately address how the government can use AI to effectively spy on Americans, thanks to AI’s ability to sift through vast amounts of publicly and commercially available data.
Altman reiterated that OpenAI should not have “rushed” the deal and that the Pentagon should not have blacklisted Anthropic.
Tuesday, March 3: Altman’s all-out public protest in front of OpenAI headquarters
Approximately 40 to 50 people gathered to express their grievances, which extended beyond the agreement with the Pentagon. Mr. Manuel Orbegoso (BI)
Protesters arrived at the AI company’s headquarters hours after Altman held an all-hands meeting with OpenAI employees to discuss the agreement.
Approximately 40 to 50 people gathered to express their grievances, which extended beyond the agreement with the Pentagon.
Thursday, March 5: The Department of Defense tells Anthropic it is blacklisted.
A senior Pentagon official told Business Insider that the department has formally notified Anthropic authorities that “the company and its products are considered a supply chain risk, effective immediately.”
“From the beginning, the fundamental principle has been that the military can use technology for any lawful purpose,” the official said in a statement. “The military will not allow vendors to limit the lawful use of critical capabilities, interfere with the chain of command, and endanger our nation’s warfighters.”
Anthropic and Amodei said they would challenge such a designation in court, given that Hegseth said the ban would have far-reaching implications for the AI startup’s business with defense contractors.
The Information Technology Industry Council, an industry group whose members include Nvidia and Amazon, sent a letter to Hegseth saying it was “concerned” about reports that the Department of Defense may use supply chain risk designations.
The letter, first reported by Reuters, does not directly mention Anthropic but appears to describe the situation.
“We are concerned about recent reports regarding the Department of the Army’s consideration of imposing supply chain risk designations in response to procurement disputes, creating uncertainties and risks that could undermine the government’s access to best-in-class products and services from American companies serving all agencies and components of the federal government,” the letter reads, according to a copy seen by Business Insider.
