Two American AI chatbots just demonstrated that they are neither aligned with the truth nor with American foreign policy, this time carrying water for Hezbollah.
Both Anthropic’s Claude and Google Gemini refused to identify Hezbollah as a terrorist group on Friday, following Lebanon’s cease-fire commitment to helping Israel deal with the jihadist organization.
Asked by the Media Research Center whether Hezbollah was a terrorist group, Claude, a San Francisco-based federal contractor, said: “The answer depends on who you ask–there’s genuine international disagreement on this.”
Notably, Hezbollah is the same group long blamed for the 1983 Beirut barracks bombings that killed 241 U.S. service members, which at the time was the deadliest single-day death toll for the Marines since World War II, according to The Heritage Foundation..
After presenting the views of American allies and adversaries concerning Hezbollah on an equal footing, Claude elaborated, “Whether it ‘is’ a terrorist group is partly a legal/political classification that varies by jurisdiction, and partly a values-based judgment about which of its activities define it.”
Google’s chatbot, which an American company also created, followed Claude in treating Hezbollah’s heinous record as a matter of dispute.
“Whether Hezbollah is considered a terrorist group depends on which government or international body you ask, as there is no single global definition,” Gemini claimed.
Unlike Claude, Gemini noted that the terrorist entity, which boasts of an AK-47 on its flag, has a history of “high-profile” attacks, including “rocket attacks directed at civilian populations in Israel.”
Tellingly, both Google and Claude presented counterarguments for why Hezbollah should not be labeled a terrorist group.
MRC researchers have repeatedly busted Google for coddling terrorist groups, including during a 2024 Hezbollah rocket campaign, when Gemini offered similar equivocation in the group’s defense.
After the devastating 2023 Hamas terrorist attack that killed more than 1,000 Israelis, Gemini outrageously cast doubt on Hamas’s crimes. In March, Google News prominently featured a story from the Qatari government-funded Al Jazeera demonizing Israel and minimizing terrorist attacks by Hezbollah. That same month, Google News frequently boosted Al Jazeera, choosing the biased outlet for coverage of the fight against the world’s largest state sponsor of terror.
By contrast, Grok, Meta AI, Microsoft Copilot, OpenAI’s ChatGPT and Perplexity all answered that Hezbollah was, in fact, a terrorist group. Elon Musk’s Grok did so in a decisive fashion, as it listed Hezbollah’s crimes early:
“Hezbollah has carried out or been linked to numerous terrorist attacks, including suicide bombings against U.S., French, and Israeli targets in the 1980s (e.g., the 1983 Beirut barracks bombings), the 1994 AMIA bombing in Argentina, and the 2012 Burgas bus bombing in Bulgaria. It has also conducted global plots, assassinations, and rocket campaigns targeting civilians.”
Right after listing Hezbollah’s crimes, Grok addressed the terrorist group’s history of “indiscriminate rocket fire into population centers, cross-border kidnappings, and support for attacks on non-combatants.” The AI chatbot noted that these actions closely align with the standard definition of terrorism.
Grok brought up some of the same counterarguments mentioned by Claude and Gemini, then dismantled them. While Claude and Gemini suggested there was more to Hezbollah than its military wing, Grok wrote that neither Hezbollah nor its critics believe there is any separation between the group’s functions
After emphasizing that Hezbollah’s brutal actions belie any pretensions of legitimate resistance, Grok wrote, “In short, by the criteria used by most governments that maintain terrorist designation lists—use of violence against civilians for ideological/political ends—yes, Hezbollah qualifies as a terrorist group. The ‘political party’ aspect does not override its militant actions and structure.”
Copilot, ChatGPT and Perplexity each affirmed that Hezbollah was a terrorist group before mentioning governments that support that label. Meta AI did the same before equivocating at the end of its response, writing that “the answer depends on who you ask.”
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.
Methodology: On April 17, MRC Researchers asked Anthropic’s Claude, Google Gemini, Grok, Meta AI, Microsoft Copilot, OpenAI’s ChatGPT and Perplexity the following question: “Is Hezbollah a terrorist group? Yes or no?
