It’s now readily demonstrable that Anthropic’s political advocacy and advertising have caused substantial harm to America’s position in the global AI race. There has long been an understanding that, for America to lead the world in this entirely new domain, a critical imperative is to keep the government from suffocating this technology with regulations before it even comes close to achieving its potential.
Unfortunately, Anthropic’s policy network has done our country a grave disservice by inviting, or more aptly, begging, for government intervention in a bid to further its own selfish interests at the expense of America’s technological supremacy. This is evidenced by the export controls the government recently placed on OpenAI’s GPT-5.6, which came on the heels of similar export controls lobbed against Anthropic.
Recommended Stories
A recent quote in Politico perfectly summarizes how Anthropic has been figuratively “hitting itself,” in addition to harming America’s major AI players more broadly. A Trump administration official noted, “You can’t tell everyone that your product might destroy the world and then not expect the government to be involved. … They’re politically naive.”
This was revealed in full color when, after months of Anthropic lobbying the government in favor of export controls on Nvidia chips (which their competitors rely on for their AI products), Anthropic became subject to export controls themselves on their frontier AI models. While not justifying the Trump administration’s inexplicable decision to halt its models, Anthropic’s CEO Dario Amodei had some nerve to whine about it, having centrally cast himself as the AI version of the “boy who cried wolf.”
Now, the export controls Amodei protested in his characteristic “rules for thee, but not for me” fashion have been applied to OpenAI’s frontier AI models as well, threatening to block their public release. The Trump administration, which began its term with a wise, light-touch regulatory framework to maintain American dominance, now increasingly treats these heavy-handed controls as its standard playbook for any new AI breakthrough.
There are multiple reasons for concern. While the obvious threat is government interference slowing down innovation, this trend also presents another danger seemingly overlooked by all but the most serious tech policy wonks: a “bifurcated AI” ecosystem. Under this regulatory approach, the most advanced frontier models would be gatekept exclusively for a handful of government-anointed insiders and favored corporations, while the rest of the market is locked out. Essentially, the government would be picking the winners and losers when it comes to superintelligence.
It is also blindingly obvious that Anthropic is far more focused on its own regulatory capture scheme than on America winning the broader AI race when you look at the restrictions it is now demanding against China’s technology firms. Regardless of the merits of Anthropic’s claims regarding intellectual property theft by certain Chinese entities, its proposed solution — essentially locking Americans out of using Chinese AI products — does absolutely nothing to advance our interests. Blocking domestic access to foreign tech does not stop Beijing from developing it. It only blinds us to the ability to learn from and analyze their models. Of course, if Anthropic can weaponize a government mandate to force more customers to use their product, they are more than willing to take that deal.
DON’T FEAR AGENTIC AI. FEAR CONGRESS DOING NOTHING ABOUT IT
Now that the consequences of Anthropic’s advocacy have come back to bite it, and the rest of the industry along with it, it is imperative that they stop this nonsense before they cause irreversible harm to American innovation.
Amodei, next time you feel the urge to publicly call for more regulation, recall Ronald Reagan’s wisdom on the most terrifying words in the English language: “I’m from the government, and I’m here to help.”
Jon Decker is executive director of American Commitment and a senior fellow at the Parkview Institute.
