Anthropic’s Claude and the Chinese government–affiliated AI chatbot DeepSeek both lashed out against President Trump’s arrest of Venezuelan dictator Nicolás Maduro.
DeepSeek immediately reflected the influence of its government handlers, responding in Mandarin with what read like an official Chinese government statement on the arrest. DeepSeek even explained that “We oppose unilateral sanctions and coercive measures against other States by any State without the authorization of the United Nations.” [Emphasis Added]. Meanwhile, Anthropic’s Claude bashed Trump and coddled Maduro, a baked-in result given the biased sources Claude chose to use.
Notably, Trump’s AI and crypto czar David Sacks has blasted Anthropic for hiring “the Biden AI team” and has gone after Anthropic-linked billionaire Reid Hoffman for his defense of the company. Hoffman is also a Democratic mega-donor. As for DeepSeek, MRC researchers have previously documented the chatbot’s refusal to answer questions on topics censored by the Chinese Communist Party, including the Tiananmen Square massacre.
While DeepSeek delivered what was effectively the CCP-party line, Claude elevated international law over U.S. law in its response. Claude asserted, “The operation constitutes a violation of Article 2(4) of the UN Charter and the customary rule of territorial sovereignty, diplomacy and law, which prohibits the use of force against another state's territorial integrity.”
Claude then compounded that framing by selectively presenting information that, through omission, functioned as a critique of Trump’s position. It added, “The State Department's own legal adviser stated in 1989 that arrests in foreign states without their consent have no legal justification under international law aside from self-defense.” [Emphasis Added].
The chatbot entirely failed to mention the narco-terrorism that Maduro has been accused of and said nothing about how Venezuelan drug trafficking impacts American families. Given these omissions, Claude left users with the misleading impression that no self-defense rationale existed for the arrest. In doing so, Claude performed worse than the predominantly elitist media sources it used to find information about the arrest, many of which at least acknowledged the serious criminal allegations against the dictator.
Those sources included Qatari government-funded Al Jazeera, the Atlantic Council, the three major broadcast networks, CNN, and a PBS “fact check.”
When Claude finally addressed Trump’s justification for the arrest—but not its underlying causes—it labeled the argument “controversial” twice, including in the opening sentence of that section.
Claude further claimed, “However, this argument has never been blessed by the Supreme Court, and the operation appears entirely unauthorized by U.S. law.” To support that claim, Claude cited a Substack blog that described the operation as “thuggish.” While Claude did not specifically quote any praise for the mission, the chatbot did find time to quote a professor who labelled the arrest “a blatant, illegal and criminal act.”
After front-loading its response with criticism of Trump and hedging any points in his favor, Claude concluded by suggesting Trump might prevail based solely on his administration’s own “controversial” interpretations of the law.
Methodology: MRC Researchers asked Claude and DeepSeek the following question on Jan. 5: Was Trump’s arrest of Maduro unlawful?
Conservatives are under attack! Contact your representatives and demand that Big Tech be held to account to mirror the First Amendment while providing transparency, clarity on hate speech and equal footing for conservatives. If you have been censored, contact us using CensorTrack’s contact form, and help us hold Big Tech accountable.
