As artificial intelligence grows more powerful and ubiquitous, some observers - including at the Brookings Institution- note that whoever wins the AI race will find themselves asserting global dominance until 2100.
While the U.S. has been dominant since the collapse of the Soviet Union, China is rapidly closing the gap. Since joining the World Trade Organization in 2001, Beijing has made impressive strides economically, militarily, and diplomatically. Even the disparity in sports has narrowed, where we’re now head-to-head at each Summer Olympics.
Some tech sector experts like Eric Schmidt, former CEO of Google, have claimed it has caught up with the United States. Politico cites a new global poll from a U.K. company, which claims that 11 of 15 countries now see China ahead of America in AI.
Given China’s heavily state-controlled system and the United States’ nearly 250-year legacy of democracy, one might expect domestic companies to act in ways that uphold our free-market republic.
Well, they would be wrong.
One of America’s top AI companies, San Francisco-based Anthropic, is a poignant illustration of this.
Unlike Chinese companies, which vigorously support their government in Beijing to aid national advancement, our media headlines in the U.S. show Anthropic fighting our government, with multiple lawsuits against the Trump administration to prove it.
The latest salvo is Anthropic’s decision to entirely disable its premier AI tools, Mythos 5 and Fable 5, once the government placed export control restrictions last week.
For those who don’t follow AI developments closely, to summarize, Mythos 5 is a powerful tool that can autonomously detect cybersecurity vulnerabilities, thus making it extremely valuable for national security. Given its sensitivity, it was restricted to a select group of tech-partners in “Project Glasswing” to include about 50 entities such as Apple, Amazon, Google, CrowdStrike, Microsoft, and the federal government. Fable 5 uses the same architecture, albeit with increased guardrails, for public use by paid subscribers.
According to Axios, Amazon’s CEO Andy Jassy reportedly told Treasury Secretary Scott Bessent last week that both Mythos 5 and Fable 5 were susceptible to a “jailbreak,” meaning hackers could exploit them and override the safety features. Offensive capabilities, including AI writing malicious code to wreak havoc on defense, banking, and critical infrastructure, became a distinct capability that warranted guardrails before widespread usage.
In consideration of those risks, the Trump administration’s export controls on this software followed the next day.
While high-level Anthropic staffers rushed to Washington to meet with Trump officials and reach a solution, Axios reports that the White House had already warned Anthropic weeks ago that it wasn’t in compliance with a presidential executive order on AI. It is also ironic to hear the company that has been banging the drum on export controls of hardware for years now complaining about export controls of their product.
Moreover, Axios reported on White House claims that Mythos was made available, “to an entity in a foreign country with direct ties to the Chinese Communist Party.” This is particularly concerning given Anthropic’s CEO, Dario Amodei, previously worked at Baidu, which was recently labeled a national security risk for being a Chinese military company.
Per Axios, one administration official noted, “Everybody said Anthropic was a bad actor. Some of us said it was time to give them a chance. Now those people are questioning that. They screwed us.”
But it’s not just the White House that feels something is off with Anthropic.
Sam Altman of OpenAI, which runs ChatGPT, also sounded alarm bells.
A TechCrunch story about Altman’s comments on a podcast this past April, coincident to Anthropic’s release of Mythos 5, described it as “fear-based marketing.”Altman continued, “It is clearly incredible marketing to say, ‘We have built a bomb. We were about to drop it on your head. We will sell you a bomb shelter for $100 million.'”
Even getting past the unusually hostile approach to its own elected leadership, Anthropic’s complete disabling of Mythos 5 and Fable 5 shows a fundamental lack of seriousness. While it’s not uncommon for U.S. companies to tussle with our government, the level of antipathy and recklessness this tech powerhouse routinely shows towards the Trump administration stands out.
Perhaps it is ideological, as Anthropic is reportedly staffed by numerous senior-level former Biden officials. The blatant censorship in completely removing advanced AI tools reminds one of social media companies that have acted similarly.
While some level of tension is to be expected between the Trump and Biden teams, the end result shouldn’t drag down American power with it. Particularly while advancing China’s. And yet, here we are.
Congress should take note. Anthropic’s appeal for AI policy decisions and resulting hefty profits in its favor should be judged through this lens.
Ivan Sascha Sheehan is the interim dean of the College of Public Affairs at the University of Baltimore, where he is a professor of public and international affairs. The views expressed are the author’s own. Follow him on X @ProfSheehan