Trump White House Prepares Executive Order to Remove Anthropic AI From Federal Systems Amid Pentagon ‘Supply Chain Risk’ Blacklist Fight
The White House is reportedly preparing an executive order to force federal agencies to rip Anthropic’s Claude out of government systems as part of an escalation in which the Pentagon has formally labeled Anthropic a “supply‑chain risk” after the company refused to lift guardrails prohibiting mass domestic surveillance and fully autonomous weapons. Claude — the only commercial large‑language model on U.S. classified networks and used in operations including the Maduro capture and recent strikes — now faces a federal ban and contract terminations, prompting Anthropic to file lawsuits alleging unlawful retaliation and constitutional and administrative‑law violations as other AI firms negotiate separate classified‑access arrangements.
📌 Key Facts
- The Pentagon pressed Anthropic (and other AI labs) to agree that their models could be used for “all lawful purposes”; Anthropic refused to remove two explicit guardrails — bans on mass domestic surveillance of Americans and on powering fully autonomous lethal weapons — and framed those limits as protecting U.S. values and safety.
- Claude is currently the only large‑language model authorized on U.S. classified networks and has been used operationally (including in an Anthropic–Palantir–supported operation to capture Nicolás Maduro and in strikes related to Iran), making any removal disruptive for military and sensitive national‑security workflows.
- Defense Secretary Pete Hegseth and DoD officials issued an ultimatum (pegged to a 5:01 p.m. deadline in public reporting), publicly attacked Anthropic leadership, and the Pentagon formally designated Anthropic a “supply chain risk,” barring military contractors from commercial activity with the firm; President Trump directed “EVERY Federal Agency” to cease using Anthropic technology and allowed a roughly six‑month wind‑down while threatening enforcement.
- The White House was reported to be preparing an executive order to force agencies to “rip out” Anthropic’s AI from federal systems; some agencies (including Treasury) and prime contractors (e.g., Lockheed, Boeing) were already asked to inventory and begin offboarding Claude ahead of formal action.
- Anthropic filed two lawsuits (in the Northern District of California and the D.C. Circuit) seeking to vacate and block the supply‑chain‑risk designation, arguing it is legally unsound and constitutes unconstitutional retaliation for the company’s protected speech (First Amendment, APA and due‑process claims).
- Industry responses were mixed but consequential: OpenAI struck a Pentagon deal after Anthropic refused and said it shares Anthropic’s red lines; other vendors (and Microsoft lawyers) expect to continue non‑defense work with Anthropic and characterise the DoD designation as legally narrow, while some labs reportedly agreed to DoD terms on classified access.
- The dispute has major business and political stakes — Anthropic has raised and attracted tens of billions in investment and projects large revenues and valuations — prompting investor concern, interventions and warnings from senators and national‑security experts about operational disruption and the unusual precedent of using a supply‑chain tool against a U.S. company.
- Despite the designation and litigation, multiple sources reported the Pentagon continued to use Claude in active operations while agencies, vendors and courts move to implement, contest, or clarify the scope of the ban.
📊 Relevant Data
In facial recognition technology, the error rate for identifying darker-skinned women is up to 34.7%, compared to 0.8% for light-skinned men, based on studies highlighting persistent racial biases.
Biased Technology: The Automated Discrimination of Facial Recognition — ACLU Minnesota
A 2026 poll found that 79% of Americans believe a human should always make the final decision before any use of lethal force by AI, with similar levels across Democrats (81%) and Republicans (81%), but no specific racial breakdowns provided in the poll.
Americans Want Humans in Control of AI — Information Technology and Innovation Foundation (ITIF)
Black service members are less likely to become officers and, as a result, are more likely to be seriously injured serving their country than their White counterparts, according to 2020 military data analysis.
Military data reveals dangerous reality for black service members and veterans — CNN
AI systems in conflict zones can perpetuate racial discrimination through targeted surveillance and over-policing of communities of African descent, as reported in 2025 analyses of law enforcement AI use.
Impacts of Artificial Intelligence (AI) on People of African Descent — Geneva International Centre for Justice
Historical precedents for US executive orders targeting companies for national security primarily involve foreign entities like Huawei and TikTok, with little direct naming of domestic US companies in such orders from 2016-2026.
Tracking regulatory changes in the second Trump administration — Brookings Institution
📊 Analysis & Commentary (7)
You are no longer the smartest type of thing on Earth
"The piece is an opinion/deep dive linking recent demonstrations of advanced AI capabilities—exemplified by Anthropic’s Claude finding hundreds of zero‑day bugs—to the conclusion that AI has become a new dominant cognitive force, creating acute dual‑use and governance challenges that demand urgent, pragmatic policy and technical responses."
Updated thoughts on AI risk
"A reasoned commentary prompted by reports that Anthropic’s Claude autonomously found hundreds of zero‑day bugs, arguing that these concrete dual‑use capabilities make AI risk urgent and require layered technical controls, access governance, coordinated disclosure/defense and policy fixes rather than mere sloganeering."
The Pentagon Threatens Anthropic
"The piece critiques the Pentagon’s tactic of threatening Anthropic with being labeled a "supply‑chain risk" to force removal of safety limits, arguing procurement leverage is a dangerous substitute for transparent legal and democratic oversight and that coercing labs to enable mass surveillance or fully autonomous weapons will erode safety and public trust."
"All Lawful Use": Much More Than You Wanted To Know
"A detailed, critical deep‑dive arguing that the administration’s demand to let Anthropic’s Claude be used for 'all lawful purposes' is vague, coercive, and dangerous — likely to strip essential safety guardrails, disrupt classified programs, and should be replaced by narrowly scoped, auditable exceptions and oversight rather than a blanket ultimatum."
If AI is a weapon, why don't we regulate it like one?
"The piece uses the Pentagon–Anthropic standoff to argue that if advanced AI truly functions as a weapon, it should be regulated through clear, democratic, weapon‑style frameworks (licensing, export controls, international norms) rather than ad hoc labels and unilateral demands that risk politicization and harm to both civil liberties and innovation."
STEVE FORBES: The AI Cold War has begun and America cannot afford to lose
"The piece is an opinion‑level response to debates like the Pentagon’s move against Anthropic, arguing the U.S. must lead in open, exportable AI grounded in democratic values, favor light‑touch rules and competitive public‑policy support rather than ceding the global standard to China or resorting to heavy regulation."
All wired up and nowhere to go
"A critical deep‑dive arguing that the Pentagon’s designation of Anthropic as a supply‑chain risk and the presidential ban on Claude are heavy‑handed, politically fraught moves that will disrupt government capabilities, risk civil‑liberties tradeoffs, prompt litigation, and require a more nuanced, targeted approach to manage AI risks."
📰 Source Timeline (37)
Follow how coverage of this story developed over time
March 09, 2026
11:30 PM
Scoop: White House readies executive order to weed out Anthropic
New information:
- Axios reports the White House is actively preparing an executive order that would formally instruct the entire federal government to "rip out" Anthropic’s AI from operations.
- The order could be issued as soon as this week, according to a source familiar with the planning.
- Axios confirms that agencies such as the Treasury Department have already begun offboarding Anthropic even ahead of the executive order.
- Anthropic’s lawsuit explicitly argues that Congress, in its procurement statutes, never gave the administration authority to blacklist a U.S. company over what the firm calls protected speech (i.e., its AI safeguards).
- The administration is characterizing Anthropic’s safety “safeguards” as a national security threat in the context of industry intervention during military operations.
- A White House official declined to confirm the EO, saying any policy announcement would come from the president and calling current discussion 'speculation,' underscoring the administration’s public posture.
7:56 PM
Anthropic accuses Trump administration of ‘pure retaliation’ in deeming the AI company a risk
New information:
- Confirms the California complaint was filed Monday, March 9, 2026, in the U.S. District Court for the Northern District of California and assigned to Judge Rita Lin, a Biden appointee.
- Details Trump’s Feb. 27 directive ordering all federal agencies to 'IMMEDIATELY CEASE all use of Anthropic’s technology.'
- Quotes Defense Secretary Pete Hegseth’s order designating Anthropic a 'Supply‑Chain Risk to National Security' and barring military contractors, suppliers or partners that do business with the U.S. military from conducting any commercial activity with Anthropic.
- Clarifies that other federal agencies followed the Pentagon’s supply‑chain risk designation with their own actions against Anthropic.
- Notes Anthropic has also filed a separate petition in the D.C. federal appeals court seeking review of Hegseth’s directive.
- Includes Anthropic’s characterization of the actions as 'pure retaliation' for its stance that Claude cannot safely or reliably be used for autonomous lethal warfare or mass surveillance of Americans, and its assertion of First Amendment and due‑process violations.
- States that the lawsuit explicitly alleges violations of the Administrative Procedure Act in addition to constitutional claims.
5:30 PM
Anthropic sues the Trump administration over 'supply chain risk' label
New information:
- Confirms two separate lawsuits were filed March 9, 2026: one in the U.S. District Court for the Northern District of California and one in the D.C. Circuit Court of Appeals.
- Details Anthropic’s core First Amendment claim that the Pentagon and Trump administration retaliated against the company for its 'protected viewpoint' on AI safety and guardrails around lethal autonomy and mass surveillance.
- Adds specific language from the complaint stating that officials are 'seeking to destroy the economic value created by one of the world's fastest-growing private companies.'
- Provides more granularity on Anthropic’s stated red lines: no use of Claude for lethal autonomous weapons without human oversight and no mass surveillance of U.S. persons.
- Clarifies that Pentagon officials publicly dispute that the conflict is about lethal autonomy or mass surveillance and argue that private firms cannot dictate 'lawful' military uses.
- Notes expert assessment that applying the 'supply chain risk' label to a U.S. frontier AI company — historically used for foreign adversary contractors — is 'highly unusual.'
- Specifies that Trump’s follow‑on social media directive ordered all federal agencies, not just the Pentagon, to stop using Anthropic’s tools.
- Reiterates that Anthropic had been the first frontier AI lab cleared for use on classified U.S. government networks before being blacklisted.
4:18 PM
Anthropic sues in federal court to reverse Trump administration's 'supply chain risk' designation
New information:
- Confirms that Anthropic filed two separate lawsuits: one in a California federal district court and another in the D.C. federal appeals court, each targeting different aspects of the Pentagon’s actions.
- Spells out the specific guardrails Anthropic refused to lift: banning mass surveillance of Americans and fully autonomous weapons, versus the Pentagon’s demand for 'all lawful uses' of Claude.
- Clarifies that the Pentagon’s 'supply chain risk' authority was originally designed to block foreign adversaries from compromising U.S. national‑security systems, and this is the first known use against a U.S. company.
- Provides updated business scale figures: Anthropic projects about $14 billion in revenue this year, has more than 500 customers paying at least $1 million annually, and was recently valued at roughly $380 billion.
- Adds direct language from Anthropic’s legal filings accusing the administration of unconstitutional retaliation for protected speech and asserting that no federal statute authorizes the designation in this context.
- Notes that the lawsuits also name the Treasury and State Departments for ordering employees to stop using Anthropic’s services, extending the dispute beyond the Pentagon.
4:04 PM
Anthropic sues Pentagon over "supply chain risk" designation
New information:
- Details that Claude is currently the only AI model authorized for use on classified networks and that the Pentagon has continued using it during the U.S. and Israeli war with Iran.
- Specific description of the guardrails Anthropic sought: no mass surveillance of U.S. citizens and no use to power lethal autonomous weapons, versus the Pentagon’s insistence on access for 'all lawful use.'
- Timeline and language of Trump’s order directing all federal agencies to 'IMMEDIATELY CEASE all use of Anthropic's technology' and Hegseth’s plan to phase Anthropic out of defense contracts over six months.
- Anthropic’s characterization of the Pentagon and Trump actions as an unconstitutional retaliation campaign targeting its 'protected speech,' and its claim that contracts are already being canceled and hundreds of millions of dollars are at risk.
- Confirmation that Hegseth has now formally issued the supply chain risk designation, which Anthropic is asking the court to declare 'arbitrary, capricious, an abuse of discretion, and contrary to law.'
3:25 PM
Anthropic sues Pentagon over rare "supply chain risk" label
New information:
- Anthropic filed two lawsuits on Monday: a primary suit in the U.S. District Court for the Northern District of California and a second, narrower challenge in the D.C. Circuit Court of Appeals.
- The complaints argue that the Pentagon’s 'supply chain risk' designation under 10 U.S.C. 3252 violates Anthropic’s First Amendment rights by punishing it for publicly opposing mass domestic surveillance and fully autonomous weapons uses of its AI.
- Anthropic contends Congress required the Pentagon to use the 'least restrictive means' to mitigate supply‑chain risk and that procurement statutes do not authorize blacklisting a domestic supplier over policy disagreements.
- The company is asking courts to vacate the designation, block its enforcement, and require federal agencies to rescind directives to drop Anthropic’s tools.
- The Pentagon responds that the dispute is about operational control and the military’s ability to use technology legally, not about speech, and maintains the designation is about keeping vendors from inserting themselves into the chain of command.
- Despite the litigation, Anthropic says it remains committed to serving the Pentagon during active combat operations and is open to resolving the dispute through dialogue.
- Microsoft and Google are expected to continue non‑defense work with Anthropic even as defense‑related federal use is curtailed.
March 07, 2026
2:11 AM
Pentagon's chief tech officer says he clashed with AI company Anthropic over autonomous warfare
New information:
- Defense Undersecretary and Pentagon chief technology officer Emil Michael says the dispute with Anthropic centered on whether Claude could be used in fully autonomous weapons for President Trump’s proposed 'Golden Dome' space‑based missile defense program.
- Michael describes wanting a 'reliable, steady partner' that will 'work with me on autonomous' systems for swarms of armed drones, underwater vehicles and other platforms, and calls Anthropic’s ethical limits on fully autonomous weapons 'irrational' from his perspective.
- He outlines specific scenarios in which the Pentagon wants AI autonomy — such as a 90‑second window to respond to a Chinese hypersonic missile and autonomous lasers knocking down drones over sleeping U.S. troops — and frames these as low‑risk uses.
- Michael recounts months‑long talks with Anthropic CEO Dario Amodei and says he concluded the company might 'wig out in the middle' of future autonomous‑weapon projects, prompting him to back the supply‑chain‑risk designation.
- The article notes that Anthropic says it sought only to bar two uses — mass surveillance of Americans and fully autonomous weapons — and that it intends to sue over the Pentagon’s designation.
March 06, 2026
6:08 AM
Pentagon labels AI company Anthropic a supply chain risk
New information:
- Pentagon statement says it has 'officially informed Anthropic leadership the company and its products are deemed a supply chain risk, effective immediately,' signaling negotiations are effectively over.
- Anthropic CEO Dario Amodei says the company will challenge the designation in court, arguing it is not legally sound and that its guardrails only target mass domestic surveillance and fully autonomous weapons at a high‑level usage layer.
- Trump has given the military six months to phase out Claude, which is already widely embedded in military and national‑security platforms.
- Lockheed Martin says it will follow the President’s and the 'Department of War’s' direction, is already looking to other large‑language‑model providers, and expects 'minimal impacts' because it is not dependent on a single vendor.
- Microsoft’s lawyers concluded that the designation only applies when Claude is used as a direct part of military contracts, allowing Microsoft to continue working with Anthropic on non‑defense projects.
- Sen. Kirsten Gillibrand criticizes the move as 'a dangerous misuse of a tool meant to address adversary‑controlled technology,' highlighting concern that a foreign‑adversary supply‑chain rule is being repurposed against a domestic firm.
1:08 AM
Scoop: Anthropic CEO apologizes for leaked memo criticizing Trump
New information:
- Dario Amodei published a blog‑style statement apologizing for the 'tone' of a leaked internal memo criticizing the Trump administration, saying it did not reflect his 'careful or considered views' and was written six days earlier.
- Amodei emphasized that Anthropic did not leak the memo and said the company’s 'most important' goal is ensuring U.S. warfighters and national‑security officials are not deprived of Claude in the middle of a war.
- A senior Pentagon official told Axios that the Department of War has officially informed Anthropic leadership that the company and its products are now deemed a supply‑chain risk 'effective immediately,' and said the military will not allow a vendor to 'insert itself into the chain of command' by restricting lawful uses.
- Microsoft stated its lawyers view the Pentagon’s designation as narrow and believe Anthropic products, including Claude, can remain available to customers other than the Department of War through platforms like M365, GitHub and Microsoft’s AI Foundry, and that Microsoft can continue non‑defense work with Anthropic.
- Despite the formal designation, a source familiar said the Pentagon was still actively using Claude on Thursday night to support military operations, including in Iran.
- The article notes that the Pentagon’s deadline for Anthropic to accept its 'all lawful purposes' standard passed without an immediate designation announcement, and that OpenAI’s subsequent deal with the Pentagon drew criticism over protections on mass domestic surveillance and autonomous weapons, prompting Sam Altman to strengthen language in that agreement.
March 05, 2026
9:35 PM
Pentagon formally designates Anthropic a supply chain risk amid feud over AI guardrails
New information:
- A senior Pentagon official confirms the Defense Department has now formally designated Anthropic a 'supply chain risk,' a move that could broadly cut it off from military contracts.
- Anthropic learned of the formal designation on Thursday; a source says no specific timeline for off‑boarding Claude was provided in that notice, despite Hegseth’s earlier public talk of a six‑month phase‑out.
- CBS reports the U.S. military has used Anthropic’s Claude model in its strikes on Iran that began last weekend, though the exact operational role remains unclear.
- Pentagon CTO Emil Michael is quoted arguing that mass surveillance of Americans and fully autonomous weapons are already illegal or restricted by internal policy, saying 'at some level, you have to trust your military to do the right thing' and that DoD will not promise in writing it will refrain from certain uses 'to a company.'
- Anthropic CEO Dario Amodei reiterates the firm’s two non‑negotiable red lines — no use of Claude for mass surveillance of Americans and no fully autonomous weapons that target people — and says the company will not move on them, arguing current law lags behind AI capabilities.
- The Pentagon offered a written compromise acknowledging existing laws and policies restricting mass surveillance and autonomous weapons, but Anthropic rejected it as 'paired with legalese' that would effectively let the military disregard practical guardrails.
- Trump administration officials escalate rhetoric, with Defense Secretary Pete Hegseth calling Anthropic 'sanctimonious,' Emil Michael describing Amodei as having a 'God‑complex,' and President Trump branding the company 'radical left' and 'woke.'
- Anthropic has previously vowed to legally challenge any supply‑chain‑risk designation, calling it 'legally unsound' and a 'dangerous precedent' for any U.S. firm that tries to negotiate guardrails with the government.
March 04, 2026
10:08 PM
Scoop: White House casts doubt on Pentagon-Anthropic reconciliation
New information:
- Axios, citing an administration official, reports that Dario Amodei’s internal comments to staff disparaging President Trump could 'blow up' chances of resolving the Pentagon–Anthropic dispute.
- The Information previously reported Amodei sent a memo to staff saying Trump disliked Anthropic for not giving him 'dictator‑style praise' and calling the OpenAI–Pentagon arrangement 'safety theater.'
- An unnamed administration official told Axios, 'you can't trust Claude isn't secretly carrying out Dario's agenda in a classified setting,' framing the dispute in explicitly political‑trust terms.
- Sources say Anthropic executives have told the Pentagon they do not want operational control over how Claude is used and regret that this point 'wasn't captured well' in media coverage.
- Another source familiar with the talks tells Axios that there had been progress toward a resolution in the past couple of days, and it is unclear how much the leaked memo will derail those talks.
8:59 PM
Anthropic CEO: We're trying to "deescalate" Pentagon AI standoff to reach agreement
New information:
- Dario Amodei told investors at the Morgan Stanley TMT Conference that Anthropic is still in talks with the Pentagon 'to try to deescalate the situation' and believes the two sides 'have much more in common than we have differences.'
- A source familiar says that in the five days since Trump canceled Anthropic’s government contracts, Anthropic executives have expressed regret to Pentagon officials over what they characterize as a misunderstanding about the company’s role in military action.
- Amodei emphasized that Anthropic 'has never questioned specific military operations' and 'does not see itself as having an operational role,' while reiterating that its red lines are mass domestic surveillance and fully autonomous weapons.
- Pentagon CTO Emil Michael told CBS DoD provided written assurances referencing laws and policies that already restrict mass surveillance and autonomous weapons, while Anthropic claims that language was paired with 'legalese' that would let the military ignore those limits.
- The article reconfirms, via two sources familiar with military use of AI, that the U.S. used Anthropic’s Claude system in the attack on Iran.
10:00 AM
Anthropic ban may threaten the military's AI advantage over China
New information:
- Axios reports that Anthropic’s AI tools have now been used operationally in two very different campaigns: the Venezuela operation that led to Nicolás Maduro’s capture and ongoing Iran operations.
- DoD CTO Emil Michael says he had a 'holy cow' moment on seeing that prior‑administration Anthropic contracts contained 'dozens of restrictions' while Claude was already 'baked into some of the most sensitive and important places in the U.S. military, where we do exercise combat power.'
- Defense Secretary Pete Hegseth has publicly called the decision to blacklist Anthropic 'final' and said the relationship is 'permanently altered,' but as of Tuesday afternoon no formal supply‑chain designation had actually been issued and the Pentagon was still using Claude in Iran.
- Former Pentagon official Michael Horowitz warns that if the dispute makes it harder for the U.S. military to access cutting‑edge AI, it could undermine America’s traditional advantage in real‑world operational experience over China.
- Steven Feldstein of the Carnegie Endowment argues that, given how early the tech is, DoD can likely swap in alternative models without losing its overall AI edge, though he acknowledges disruption from ripping out Anthropic.
March 02, 2026
4:55 PM
Anthropic vs. White House puts $60 billion at risk
New information:
- Anthropic has raised more than $60 billion from over 200 venture investors, with roughly half of that coming in just the past month, greatly increasing the financial stakes of a prolonged break with the U.S. government.
- Axios reports that after Anthropic refused to relax its guardrails, OpenAI stepped in and signed the Pentagon deal that Anthropic would not, filling the military AI slot and undercutting Anthropic’s leverage.
- Defense officials were reportedly infuriated by Dario Amodei’s public blog post and saw it as 'virtue signaling' to Anthropic employees and rival AI engineers, fearing it could sour future government–AI industry engagement.
- Axios notes that Hegseth’s public tweet barring any military contractor from doing business with Anthropic goes well beyond what a formal 'supply chain risk' designation would normally require, potentially implicating companies like Nvidia that sell critical chips to Anthropic.
- DoD has not yet served Anthropic with formal notice of the supply‑chain‑risk designation; Anthropic intends to challenge it in court once it is issued, and the eventual language may be narrower than Hegseth’s threat.
- The piece highlights that venture capitalists had historically avoided deep government‑dependent tech because of political volatility and suggests this showdown may validate those concerns, with Anthropic now 'under siege from its own government.'
March 01, 2026
3:29 PM
AI executive Dario Amodei on the red lines Anthropic would not cross
New information:
- On‑the‑record network interview in which CEO Dario Amodei explicitly restates Anthropic’s two 'red lines': refusing mass surveillance of Americans and refusing to power fully autonomous weapons with no human in the loop.
- Amodei characterizes the Pentagon 'supply chain risk' designation and federal phase‑out as 'retaliatory and punitive' and 'unprecedented,' and says 'I don't know what else to call it' besides retaliatory and punitive.
- He confirms Anthropic plans legal action, saying that so far 'all we've seen are tweets from the president and tweets from Secretary Hegseth' and that they remain at the negotiating table.
- Publicly asserts Anthropic’s position that its models are not yet reliable enough to safely power lethal autonomous weapons and that the company does not want to sell something that could get U.S. troops or civilians killed.
- Responds to Trump’s 'left‑wing woke company' label by insisting Anthropic has been 'studiously even‑handed' and tried to remain non‑partisan.
3:20 PM
AI company Anthropic's Dario Amodei: "We are patriots"
New information:
- CBS interview provides on‑camera framing from Dario Amodei that Anthropic 'refused to allow' Claude to be used to power 'fully‑autonomous weapons without any human involvement' and large‑scale domestic surveillance, spelling out the specific guardrails at issue.
- Amodei characterizes the Trump administration’s Pentagon 'supply chain risk' designation and federal ban as 'retaliatory and punitive' in a national TV interview, reinforcing earlier written statements.
- He pointedly describes Anthropic as 'patriots,' pushing back on the national‑security smear implicit in the 'supply chain risk' label and arguing that refusing certain uses is consistent with U.S. interests, not hostile to them.
February 28, 2026
11:00 AM
Anthropic CEO says he's sticking to AI "red lines" despite clash with Pentagon
New information:
- Amodei tells CBS Anthropic has two explicit 'red line' guardrails: no use of Claude for mass surveillance of Americans and no use to power fully autonomous weapons that select and strike targets without human input.
- He says Anthropic will not move on those red lines but still wants to work with the Pentagon 'for the sake of U.S. national security' if the department will address those concerns.
- Pentagon CTO Emil Michael tells CBS the department wants Claude for 'all lawful purposes,' says DoD is not currently interested in mass surveillance or autonomous weapons uses, and argues existing laws and internal policies already restrict both, so written contractual carve‑outs are unnecessary.
- Amodei warns AI could enable new forms of mass surveillance via purchased private‑sector data and says current AI unpredictability makes fully autonomous weapons unreliable and raises unresolved accountability questions if they kill innocents or U.S. troops.
5:46 AM
Anthropic CEO Dario Amodei calls White House's actions "retaliatory and punitive"
New information:
- Amodei tells CBS the designation is 'retaliatory and punitive' and 'unprecedented,' and that Pentagon communications made that motive 'very clear' in their language.
- He emphasizes Anthropic sees its guardrails on mass domestic surveillance and autonomous weapons as defending 'American values' and describes the company as 'patriotic Americans' supporting U.S. national security.
- CBS notes the Pentagon awarded Anthropic a $200 million contract in July 2025 and that talks broke down over surveillance and autonomous-weapons use of Claude, with DOD rejecting Anthropic’s requested guardrails.
- The piece reiterates that Hegseth’s supply‑chain‑risk designation is the first ever issued for a U.S. company and follows Trump’s public order for all federal agencies to immediately stop using Anthropic, with up to six months to phase out at DOD.
5:35 AM
Anthropic CEO on "retaliatory and punitive" Pentagon action
New information:
- In an exclusive CBS interview, Anthropic CEO Dario Amodei publicly characterizes the Pentagon’s supply‑chain risk designation as 'retaliatory and punitive.'
- Amodei directly links the designation to Anthropic’s refusal to grant the military 'unfettered access' to its AI model, sharpening the causal claim beyond earlier background reporting.
- The interview confirms Anthropic’s position in the CEO’s own words, giving an attributable, on‑camera account that will be central in any political or legal fight over the designation.
2:29 AM
Anthropic to take Trump's Pentagon to court over AI dispute
New information:
- Anthropic issues a public statement saying it will 'challenge any supply chain risk designation in court' and calls itself 'deeply saddened' by the dispute.
- The company explicitly cites 10 U.S.C. § 3252 and argues a 'supply chain risk' label can only lawfully affect Pentagon contracts, not non‑DoD commercial work by defense contractors using Claude for other customers.
- Anthropic tells customers that all individual and commercial access to Claude remains 'completely unaffected' except for DoD‑contract work if the designation is formally adopted.
- Axios reports that the Pentagon has accepted OpenAI’s safety measures and usage limits in principle, although no contract has yet been signed.
- Anthropic reiterates it will not remove guardrails restricting uses for mass domestic surveillance or fully autonomous weapons, despite threats from Trump and Hegseth.
12:03 AM
Hegseth declares Anthropic a "supply chain risk to national security"
New information:
- Hegseth has now formally deemed Anthropic a 'supply chain risk to national security' and announced that no contractor, supplier or partner that does business with the U.S. military may conduct any commercial activity with Anthropic.
- Hegseth publicly framed the move as ensuring 'America's warfighters will never be held hostage by the ideological whims of Big Tech' and declared the decision 'final.'
- The piece details the Pentagon’s negotiating position: it demanded the right to use Anthropic’s Claude model for 'all lawful purposes,' offered to put existing legal limits on mass surveillance and fully autonomous weapons in writing, and set a 5:01 p.m. Friday deadline before pulling the plug.
- Anthropic’s response is quoted at length, with Dario Amodei arguing that Claude is not yet safe enough for fully autonomous weapons, that certain uses undermine democratic values, and that the Pentagon’s offered language was paired with 'legalese' that would let safeguards be disregarded 'at will.'
- Pentagon CTO Emil Michael is quoted saying, 'at some level, you have to trust your military to do the right thing,' and insisting the department would 'never say that we're not going to be able to defend ourselves in writing to a company.'
February 27, 2026
10:39 PM
Trump says he plans to order federal ban on Anthropic AI after company refuses Pentagon demands
New information:
- Fox piece reproduces more of Trump’s own rhetoric from his Truth Social announcement, including threats to use the 'Full Power of the Presidency' and 'major civil and criminal consequences' against Anthropic if it is not 'helpful' during the six‑month phase‑out.
- It confirms the Pentagon put a firm deadline of 5:01 p.m. ET Friday on Anthropic to accept 'any lawful use' terms before terminating the contract and branding the company a 'supply chain risk' for the Department of War.
- It carries a more detailed public rebuttal from Pentagon public‑affairs official Sean Parnell claiming DoD has 'no interest' in mass domestic surveillance or fully autonomous weapons and insisting the department only wants authority to use Anthropic’s model for 'all lawful purposes.'
10:15 PM
These federal agencies may have a Claude problem now
New information:
- Axios confirms Claude is already in active use or pilot programs at HHS, the Office of Personnel Management, Lawrence Livermore National Laboratory, Lawrence Berkeley National Laboratory, and NASA’s Jet Propulsion Laboratory.
- JPL recently completed its first‑ever AI‑planned Mars rover drive using Claude, meaning the blacklist touches live space‑operations tooling, not just back‑office pilots.
- Claude was made broadly available across all three branches of the federal government last year under a GSA OneGov agreement, expanding the scope of systems that now must be reviewed or replaced.
- Axios notes that major AI vendors priced their services cheaply to seed federal usage, and this fight may make such contracts less attractive going forward.
9:43 PM
Trump ends federal use of Anthropic's AI technology, threatens further action against company
New information:
- CBS clip reiterates that Trump has now ordered all federal agencies to stop using Anthropic’s AI technology, framed as a completed directive rather than just a threatened ban.
- It underscores that the immediate driver is a dispute with the Pentagon over military uses of Anthropic’s AI, but does not materially expand on the terms of the phase‑out or new legal steps.
9:08 PM
Trump orders federal agencies to stop using Anthropic's AI technology
New information:
- Trump publicly posted on Truth Social that he is directing 'EVERY Federal Agency' to 'IMMEDIATELY CEASE all use of Anthropic's technology.'
- He specified that agencies will have six months to phase out Anthropic’s products rather than an instant cutoff.
- Trump threatened to use the 'Full Power of the Presidency' to impose 'major civil and criminal consequences' if Anthropic does not cooperate with the government during the phase‑out.
8:53 PM
Trump moves to blacklist Anthropic AI from all government work
New information:
- President Trump publicly announced that the U.S. government will blacklist Anthropic and ordered 'EVERY Federal Agency' to immediately cease using its technology.
- The Pentagon is formally designating Anthropic as a 'supply chain risk'—a label previously reserved for firms from adversary nations such as Huawei—triggering termination of a Claude contract worth up to $200 million.
- The blacklist extends beyond the Pentagon: all federal agencies are barred from Anthropic products, with a six‑month wind‑down for agencies and vendors to offboard Claude.
- Claude is currently the only large language model on U.S. classified networks and was used in the operation to capture Nicolás Maduro; defense officials privately describe disentangling it as a 'huge pain in the ass.'
- Palantir, which uses Claude for its most sensitive military work, will be forced to re‑platform onto rival models acceptable to DoD.
- Trump framed the move ideologically, attacking Anthropic as 'radical left' and 'woke' for refusing to drop guardrails against mass domestic surveillance and fully autonomous weapons.
- Emil Michael, the senior Pentagon official steering the talks, publicly attacked Anthropic CEO Dario Amodei as a 'liar' with a 'God complex' after Amodei’s statement rejecting DoD’s 'best and final offer.'
- Anthropic has not yet said whether it will challenge the 'supply chain risk' designation in court, but faces potential loss of 'hundreds of millions of dollars' as government‑exposed customers steer clear.
8:03 PM
President Trump bans Anthropic from use in government systems
New information:
- NPR provides precise timing: Trump’s Truth Social order came about an hour before a 5:01 p.m. ET Pentagon deadline for Anthropic to drop contract prohibitions on domestic mass surveillance and fully autonomous weapons.
- The story adds detailed Trump quotes attacking Anthropic as "leftwing nut jobs" and explicitly directing "EVERY Federal Agency" to cease use of its technology.
- NPR reports new on‑record comments from OpenAI CEO Sam Altman, who publicly affirms that OpenAI shares Anthropic’s 'red lines' on banning U.S. domestic mass surveillance and entirely autonomous weapons, and says he opposes Pentagon threats to invoke the Defense Production Act.
- Altman reveals OpenAI is negotiating its own deal with the Pentagon to deploy models on classified systems with explicit exclusions for U.S. surveillance and fully autonomous weapons, underscoring that this is becoming a cross‑industry standard rather than an Anthropic one‑off.
6:08 PM
Scoop: Top Senate defense leaders intervene in Pentagon-Anthropic AI dispute
New information:
- Senate Armed Services Chair Roger Wicker, Ranking Member Jack Reed, Defense Appropriations Chair Mitch McConnell and Ranking Member Chris Coons jointly sent a private letter Friday to both Anthropic and the Pentagon urging them to resolve their dispute.
- The Senate leaders’ intervention is explicitly described as responding to growing pressure on Congress to step into the Pentagon–Anthropic standoff over use limits on Claude in classified settings.
- Axios reiterates that Pentagon officials have set a 5:01 p.m. EST Friday deadline for Anthropic to agree to 'all lawful purposes' use or risk losing its contract, and notes that Anthropic reaffirmed Thursday it will not accept the Pentagon’s demands.
3:44 PM
Dispute Between Pentagon and Anthropic Intensifies as Deadline Looms
New information:
- The New York Times piece reports that Emil Michael, the Pentagon official overseeing AI, publicly attacked Anthropic CEO Dario Amodei on X, calling him 'a liar' with a 'God‑complex' who wants to 'personally control the US Military' and is 'ok putting our nation’s safety at risk.'
- It specifies that the Pentagon’s ultimatum is pegged to a precise 5:01 p.m. Friday deadline for Anthropic to allow unrestricted military use of its most advanced model or face potential blacklisting and Defense Production Act action.
- The article notes that State Department officials have now joined the public pressure campaign on social media, reinforcing the Pentagon’s line, while Democratic senators are openly backing Anthropic’s refusal to drop safeguards.
3:22 PM
Tech company refuses Pentagon demands on unrestricted use of its AI
New information:
- Sean Parnell, Assistant to the Secretary of War for Public Affairs, publicly states the Pentagon has 'no interest' in mass domestic surveillance or fully autonomous weapons and is asking only to use Anthropic’s model for 'all lawful purposes.'
- Parnell sets a specific deadline of 5:01 PM ET on Friday for Anthropic to agree, or the Pentagon will terminate the partnership and formally deem the company a 'supply chain risk' for the Department of War.
- Under Secretary for Research and Engineering Emil Michael posts on X calling Dario Amodei 'a liar' with a 'God‑complex' and accuses Anthropic of trying to control the U.S. military.
- Michael further claims 'Anthropic is lying,' insists the department does not do mass surveillance, and frames the dispute as whether warfighters must effectively seek Amodei’s 'permission' to use AI in time‑critical scenarios like shooting down enemy drone swarms.
- War Secretary Pete Hegseth amplifies Michael’s posts by reposting them, signaling top‑level endorsement of the hard‑line stance.
- Anthropic, via statement to Fox News Digital, reiterates Amodei’s earlier position that the firm cannot in good conscience remove safeguards that bar mass domestic surveillance and fully autonomous weapons, despite Pentagon threats.
February 25, 2026
10:36 PM
Scoop: Pentagon takes first step toward blacklisting Anthropic
New information:
- The Pentagon has already contacted Boeing and Lockheed Martin to request analyses of their exposure and reliance on Anthropic’s Claude model.
- Lockheed Martin confirmed publicly that the Department of War reached out 'ahead of a potential supply chain risk declaration' about Anthropic.
- Sources say the Pentagon plans to contact 'all the traditional primes' to inventory their use of Claude before deciding whether to label Anthropic a supply-chain risk.
- Claude is currently the only AI model running in U.S. classified systems and was used in the operation to capture Venezuelan leader Nicolás Maduro via an Anthropic–Palantir partnership.
- In a tense Tuesday meeting, Defense Secretary Pete Hegseth gave Anthropic CEO Dario Amodei a deadline of 5:01 p.m. Friday to accept a Pentagon demand that Claude be available for 'all lawful purposes' or face either a Defense Production Act order or a supply‑chain‑risk designation.
- A senior Defense official told Axios the department intends to make Anthropic 'pay a price' and described any disentangling from Claude as 'an enormous pain in the ass.'
1:55 PM
Pentagon at odds with tech company Anthropic over AI model
New information:
- CBS characterizes the situation as a 'tense standoff' between the Pentagon and Anthropic over access to the company’s AI model.
- Defense Secretary Pete Hegseth has given Anthropic a specific deadline of Friday to provide the military 'full access' to an AI model, according to CBS sources.
- CBS confirms that Anthropic’s CEO is actively pushing back against the Pentagon’s demand, reinforcing that the company is resisting rather than quietly negotiating terms.
February 23, 2026
10:00 AM
Scoop: Hegseth to meet Anthropic CEO as Pentagon threatens banishment
New information:
- Defense Secretary Pete Hegseth has summoned Anthropic CEO Dario Amodei to an in‑person Pentagon meeting Tuesday morning that senior officials describe as an ultimatum rather than a 'get‑to‑know‑you' session.
- Pentagon officials say negotiations have stalled, contradicting Anthropic’s public characterization that talks have been 'productive' and 'in good faith.'
- The article details that the Pentagon is actively considering designating Anthropic a 'supply chain risk'—voiding its contracts and forcing other defense contractors to certify they are not using Claude in any Pentagon‑related workflows.
- Anthropic’s current red lines are specified: refusing to support mass surveillance of Americans and the development of weapons that fire without human involvement, even as it offers to loosen some other restrictions.
- The piece identifies who will be in the room for the Pentagon side—Hegseth, Deputy Secretary Steve Feinberg, and R&E Under Secretary Emil Michael—and quotes a senior official calling Amodei’s stance 'ideological.'
February 19, 2026
10:00 AM
Pentagon-Anthropic battle pushes other AI labs into major dilemma
New information:
- Pentagon is threatening to sever its Anthropic contract and formally declare the company a 'supply chain risk' unless it lifts model‑use restrictions and agrees to 'all lawful use.'
- Claude is currently the only large‑language model available on U.S. classified systems, accessed through Anthropic’s partnership with Palantir; OpenAI’s ChatGPT, Google’s Gemini and xAI’s Grok are used only on unclassified systems so far.
- Defense officials say at least one lab — identified by a source as Elon Musk’s xAI — has already told the Pentagon it is 'ok with all lawful use at any classification level.'
- OpenAI and Google are in active talks about classified‑domain access; OpenAI told Axios that moving into classified work would require a new or modified agreement, while Google declined comment.
- A senior official admits replacing Claude on classified networks would be 'massively disruptive,' but says the showdown with Anthropic is also being used to 'set the tone' for negotiations with the other labs.
- The Pentagon argues it cannot 'litigate individual use cases' with Anthropic before or after operations, while Anthropic remains especially concerned about enabling mass domestic surveillance and fully autonomous weapons.
February 16, 2026
7:55 PM
Maduro raid questions trigger Pentagon review of top AI firm as potential ‘supply chain risk’
New information:
- Pentagon spokesman Sean Parnell confirms the Department of War is formally 'reviewing' its relationship with Anthropic after the Maduro capture raid.
- A senior War Department official says many senior officials now view Anthropic as a potential 'supply chain risk' and are considering requiring all vendors to certify they do not use Anthropic models.
- Article specifies Anthropic holds a $200 million Pentagon contract from July 2025 and that Claude was the first commercial model brought onto classified networks.
- U.S. officials accuse Anthropic of asking whether Claude was used in the Maduro raid, allegedly signaling it might not approve such use; Anthropic denies probing specific operations and says the dispute is about its bans on fully autonomous weapons and mass domestic surveillance.
- Pentagon sources say they are pushing Anthropic, OpenAI, Google, and xAI to authorize military use of their models for 'all lawful purposes' and that other firms have already agreed to this on unclassified systems.
12:44 PM
Exclusive: Pentagon warns Anthropic will "pay a price" as feud escalates
New information:
- Pentagon is 'close' to cutting business ties with Anthropic and considering designating it a 'supply chain risk', which would effectively force DOD contractors to drop Anthropic tools.
- Senior Pentagon official told Axios they will 'make sure [Anthropic] pay a price' for resisting DOD demands over acceptable uses of Claude.
- Pentagon negotiators are insisting Anthropic, OpenAI, Google, and xAI agree that their models can be used for 'all lawful purposes', while Anthropic seeks to bar uses tied to mass surveillance of Americans and fully autonomous weapons.
- Anthropic confirms Claude is the only AI model currently available inside U.S. military classified systems and that it was used during the January raid that captured Nicolás Maduro.
- Anthropic sources warn that current U.S. surveillance law does not account for how AI can massively scale 'open source' monitoring of Americans (e.g., fusing social media, voter rolls, protest permits).
February 05, 2026