When researchers based Anthropic in 2021, they stated the race to construct highly effective AI was shifting too recklessly. They inserted detailed security measures into their merchandise and marketed their dedication to security as the company high quality that distinguished them from rivals—notably OpenAI, the rival firm they’d left. In March 2026 that popularity was examined when the Trump administration declared that Anthropic was a provide chain threat.
The corporate had refused to take away built-in safeguards that prohibited home surveillance and autonomous weapons from merchandise it had equipped to the Pentagon. President Donald Trump ordered the federal authorities to cease utilizing Anthropic and its giant language mannequin, Claude, labeling the corporate a nationwide safety threat. Inside hours, OpenAI made a deal to be the Pentagon’s provider as a substitute.
Regardless of Anthropic’s obvious stand, throughout its conflict with Trump, the corporate quietly scrapped the binding principles in its foremost security coverage. A number of weeks earlier, Anthropic’s head of safeguards analysis had resigned, warning that “the world is in peril.” And every week after the Pentagon formally banned Claude, the U.S. navy was still using the technology to pick out and goal websites to bomb in Iran.
As a thinker learning the rule of law and democracy, I’ve discovered that authoritarian governance of know-how typically doesn’t contain direct censorship. As a substitute, it delegitimizes the meant protections, poisoning any exterior regulation and even voluntary self-regulation that deviates from the regime’s targets or values.
The Trump administration, which follows the authoritarian playbook, has argued that AI security requirements and consumer restrictions are ideological impositions moderately than sound engineering choices. The ”Preventing Woke AI” executive order of July 23, 2025, didn’t change what firms are allowed to do with their merchandise. By attaching the “woke” label to fundamental ethics protections, the administration made these protections politically costly to maintain.
The Brennan Middle, a authorized coverage and advocacy group, has documented how AI ethics is being redefined through contract negotiations. In these instances, the federal government weaponizes phrases akin to biased to disqualify firms that preserve civil rights protections from competing for federal contracts.
The prisoner’s dilemma
A single U.S. Protection Division AI contract might be value billions of {dollars}. It could additionally present entry to information no personal firm might in any other case have and unlock additional authorities work. Firms that preserve the ethics guardrails threat ceding floor to rivals that don’t.
When OpenAI moved in to take the Pentagon work, CEO Sam Altman instructed his board of administrators the transfer looked “opportunistic and sloppy.” However he stated the corporate took it anyway, as a result of admitting that an motion appears dangerous is totally different from being keen to fall behind.
This case displays the basic prisoner’s dilemma. If Anthropic maintains security provisions and OpenAI strips them away, OpenAI will get the contracts and the longer term benefit. If each firms preserve the provisions, digital protections would possibly survive. However as a result of neither firm might be sure the opposite will maintain the road—and since being left behind just isn’t a superb possibility—the rational selection is to discard security measures.
These circumstances differ from a regular market race to the underside in a single key respect: The lure of getting to strip away guardrails isn’t an accident of competitors; it’s being maintained by the federal government through incentives.
Palantir didn’t wait to be caught on this lure. The info analytics firm was based by Peter Thiel and run by Alex Karp, who spent years denouncing “woke” Silicon Valley. Palantir constructed its enterprise mannequin round authorities surveillance and navy information infrastructure.
Whereas Palantir has stated it’s committed to privacy and civil liberties, critics contend that the corporate is dismantling those protections. The corporate’s inventory has surged below the Trump administration, its contracts have expanded, and it now has a front-row seat the place AI coverage is being written. Palantir solved the prisoner’s dilemma by defecting first.
It’s necessary to notice that the dissolution of safety teams across the industry, akin to OpenAI’s Superalignment crew and Microsoft’s ethics unit, isn’t the results of anybody deciding to desert security. What I see in analyzing the totally different firms’ actions is a sample: an accumulation of collective, incremental compromises that quietly reorient the definition of security away from the general public and towards the state. The ensuing hurt and dangers fall on everybody whose lives are formed by AI techniques.
Redefining security to serve the federal government
Throughout authorities contracts and coverage paperwork, I’ve additionally noticed that the original definition of AI-related safety has shifted from defending the general public towards making systems controllable for the state. The “anti-woke” framing accelerates this shift: As soon as ethics necessities are characterised as ideological moderately than technical, eradicating them might be framed not as a security discount however as a correction.
This shift doesn’t require dangerous religion from the businesses. Security groups are nonetheless doing rigorous work. The businesses aren’t mendacity after they describe their security commitments. These commitments at the moment are merely oriented towards the federal government moderately than the general public.
The case for stronger AI regulation assumes {that a} authorities constrains business entities on behalf of the general public. However blacklisting an organization for sustaining civil rights protections, after which banning the navy deployment of its AI hours later, reveals that the federal authorities on this occasion permits the hurt that regulation is supposed to forestall.
Increasing regulatory authority over AI firms doesn’t essentially defend residents. Security laws—intended to constrain corporate power—in authoritarian regimes grow to be instruments to coerce compliance.
Michael Gregory is an assistant professor of philosophy at Clemson University.
This text is republished from The Conversation below a Artistic Commons license. Learn the original article.
