Anthropic has been in the political spotlight of late. But the actual consequences of this incident on its business are not as severe as it seems, and its AI competitors will also be saddled with a lot of political risks in the future.
This large AI model research and development company is locked in a legal tug-of-war with the Trump administration. Earlier this year, the company refused to give the U.S. Department of Defense unrestricted access to its tools, which was subsequently deemed a security risk. The conflict between the two sides escalated again last month, when the U.S. government introduced coercive measures to force Anthropic to cut off external access to its most advanced models.
Although these political challenges may seem intractable, they are unlikely to persist in the long term. As the two companies prepare for IPOs, in the long run, the regulatory dilemma faced by Anthropic may even be easier to resolve than competing product OpenAI.
The essential reason why the United States introduced restrictive measures against Anthropic is that the U.S. political rule system cannot keep up with the rapid development of artificial intelligence. At this stage, the United States has very few written laws that specifically regulate the AI industry. When officials become aware of potential threats, they can only hastily introduce emergency response plans.
Some control measures appear very blunt. When the U.S. Department of Defense listed Anthropic as a supply chain risk subject, the original intention of the bill cited was to prevent foreign hostile forces from intervening in U.S. defense procurement.
Anthropic, led by CEO Dario Amodei, has launched two lawsuits seeking to overturn the risk determination. Judging from current developments, the U.S. government will be able to win at least one of the lawsuits and continue to ban Anthropic from participating in defense-related contract projects.
This ban will inevitably shrink the market scope of Anthropic's products, but being excluded from the defense procurement system will not have a major financial impact on the company. The defense business is not the company's core main business.
The bigger hidden danger for this company is that it may suffer a heavy blow to its brand reputation after being labeled as a security risk. In theory, it will lose its corporate customer base - this group contributes about 80% of Anthropic's revenue.
But looking at it from another angle, this dispute may actually enhance Anthropic’s reputation and image.
The source of Anthropic's conflict with the Department of Defense is its refusal to allow its models to be used in the development of fully autonomous weapons and large-scale surveillance. Even putting aside ethical considerations, if Anthropic is tied to the military and involved in military operation errors and public surveillance, the negative impact it will have on corporate customers will be far more fatal than being characterized as a security risk by the Pentagon.
App download data can prove that many users agree with Anthropic’s stance. According to data from the data agency Sensor Tower, after it was revealed at the beginning of this year that OpenAI was negotiating cooperation with the Pentagon, the number of uninstalls of OpenAI's ChatGPT surged significantly, even though OpenAI stated that it would set usage restrictions on the military's use of its technology. In addition, within a few days after the Pentagon listed Anthropic as a security risk, the number of new installations of Anthropic's Claude exceeded that of ChatGPT, breaking ChatGPT's consistent advantage in new users.
The U.S. Department of Defense actually has good reasons to repair its partnership with Anthropic. Anthropic has made it clear that it will not relax restrictions on the usage scenarios of its models. The Pentagon is therefore left with only two options: one is to seek continued cooperation with Anthropic, and the other is to abandon the company's technology entirely. The latter means that the Department of Defense will not have access to some of the world’s top AI models. When public opinion and policy trends change, the two sides will most likely reach a reconciliation.
Another suppression measure by the Trump administration against Anthropic has also come to an end. Last month, the United States used the export control authority of the Department of Commerce to prohibit overseas personnel (including Anthropic's own overseas employees) from using the company's technology, which directly caused Anthropic to shut down all external access to its top models Fable 5 and Mythos 5.
Now this tense situation has cooled down: the United States first allowed Anthropic to restore access to some screened and trusted customers; after Anthropic implemented network security rectifications, the government officially approved its large-scale open model access late last month.
This incident just confirms the strong technical strength of Anthropic's model, rather than the company's hard loopholes that can lead to long-term suppression. Brian Pitz, a capital markets analyst at Bank of Montreal, mentioned in a recent research report that this restriction "exactly highlights the technical strength and current industry leadership of Anthropic's large model."
From a long-term perspective, Anthropic also faces lower risk intensity than its most direct competitors.
OpenAI, like Anthropic, is advancing its IPO, with a listing valuation target expected to exceed US$1 trillion, and OpenAI is targeting a wider mass market. Almost all of Anthropic's revenue comes from corporate customers, and it firmly ranks as the market leader in the B-side market; while OpenAI covers a huge user group. Sensor Tower data shows that ChatGPT’s monthly active users exceeded 1 billion in May, far surpassing other competing products, and its user base is several times that of Anthropic.
However, OpenAI's huge user base is not only a commercial advantage, but also burdens it with heavy political baggage, requiring it to undergo all-round social scrutiny for a long time. Social media giants such as Meta have been mired in similar regulatory scrutiny for many years. OpenAI CEO Sam Altman has attended a congressional hearing on the potential risks of AI, and will inevitably be questioned many times in the future.
Altman has repeatedly proposed that the U.S. government take equity stakes in OpenAI and its competing companies. If the U.S. government invests in OpenAI but other companies do not accept the plan, OpenAI may be able to obtain certain policy preferences, but it will also be subject to administrative intervention and weaken its market competitiveness.
In contrast, Anthropic did not directly propose that the government take shares. Instead, it recently proposed the establishment of special investment accounts for Americans who have been hardest hit by AI unemployment, with shares in major AI companies held in the accounts.
Any form of government equity participation will have many hidden risks, and this is just one of the many political risks that the two companies will need to deal with in the next few years. However, there are essential differences between the two business models. Even if Anthropic is currently more focused on public opinion, in the long run OpenAI will face more prominent policy and regulatory pressures.