Al Jazeera, the virulently anti-Israel news outlet controlled by Qatar, is one of the two top sources used by leading artificial intelligence chatbots—OpenAI’s ChatGPT, Google Gemini, and Perplexity AI—to answer questions and write news summaries about the Israeli-Palestinian conflict, a Washington Free Beacon analysis has found.
ChatGPT, Gemini, and Perplexity—as well as X’s Grok—list Al Jazeera as one of their most reliable sources on the topic. In response to queries from the Free Beacon, the chatbots praised Al Jazeera for its reliability, “on-the-ground detail,” “academic credibility,” and “global visibility.”
Indeed, ChatGPT, the world’s leading AI chatbot, said that in the past month it cited Al Jazeera more frequently on the topic than almost any other news source, including the New York Times and the Associated Press. Gemini, for one, says it specifically does not use pro-Israel news sources, because they engage in “hasbara,” the Hebrew word for public relations and advocacy, “rather than journalism.” Perplexity, which reportedly boasts 22 million users, says its top sources for news on the Middle East are “respected outlets with strong regional reporting and editorial standards.” The first is Al Jazeera.
The ubiquity of Al Jazeera as a primary source for AI-generated news searches raises questions about the capacity (or willingness) of American AI giants to deliver accurate news about Israel’s war against Hamas in Gaza. This comes as AI-generated summaries are increasingly replacing traditional news publications as a primary source for news.
Since the early to mid-2000s, when it was al Qaeda’s go-to destination for taped messages from the world’s most wanted terrorists, as well as for hostage and beheading videos, Al Jazeera has earned bipartisan criticism for being a “state-controlled propaganda arm” that Qatar has used to “incite violence, glorify terrorist killers as ‘martyrs,’ and broadcast hateful, extremist content.”
The reliance of the AI tools on Al Jazeera is already raising eyebrows on Capitol Hill. “Al Jazeera is not a news organization, it is a mouthpiece for terrorists like Hamas,” Sen. Tom Cotton (R., Ark.), who had a recent run-in with an Al Jazeera correspondent, told the Free Beacon. “American AI companies should not rely on terrorist-sympathizing networks for anything, much less news about the Middle East.”
The chatbots, though, currently list Al Jazeera as one of the most reliable news sources on Middle East issues.
ChatGPT, when asked via user query, provided a “rough breakdown of which outlets [it has] drawn from most frequently in the past month when answering questions about the Israel–Palestinian conflict.”
The chatbot cited Al Jazeera and Reuters at the top of its list—as “very frequent” sources—the AP and Haaretz as “frequent” sources, and the New York Times and Times of Israel as “occasional” sources.
When asked why it relied so frequently on the Qatari-controlled news site, ChatGPT said Al Jazeera has “access on the ground,” “search visibility,” a “balance of perspectives,” and “academic/library recommendations.”
“Universities like Harvard and MIT include Al Jazeera on their curated lists of reliable conflict resources, so it’s part of the ‘standard set’ I reach for, along with Reuters, Haaretz, BBC, etc.,” said ChatGPT, using the first person singular to describe itself.
Google’s Gemini also included Al Jazeera on its list of most reliable sources on the Middle East conflict, as well as the explicitly anti-Israel outlets +972 Magazine, Mondoweiss, and B’Tselem.
Grok listed WAFA, the official news agency of the Palestinian National Authority, but included no Israeli government sources.
It defended its use of Al Jazeera as a reliable source while acknowledging its relationship with Qatar.
“Despite criticisms of bias due to its Qatar-based funding, Al Jazeera adheres to journalistic standards, employs professional reporters, and has a global reputation for covering the Middle East extensively,” the X-developed chatbot said. “It consistently provides verifiable details on events, casualty figures, and humanitarian impacts, which are cross-referenced by other outlets and organizations like the UN.”
Perplexity offered a similar defense of Al Jazeera when pressed on its Qatari funding. “Despite its funding and potential biases,” the chatbot said, “Al Jazeera is widely recognized for its extensive international reporting resources, regional perspective, and a willingness to cover stories underrepresented by Western media.”
Israel has eliminated several Al Jazeera “journalists” after finding evidence that they were also members of terrorist organizations like Hamas and the Palestinian Islamic Jihad. The Israelis have presented documentary evidence, including personnel tables, training course rosters, phone directories, salary documents, and videos, as proof that the Al Jazeera journalists were also active Hamas employees (Al Jazeera claims the evidence is fabricated).
Israel alleges (with evidence) that Anas Al-Sharif, for instance, helped lead Hamas’s missile-launching operations and belonged to the unit within the terror group that carried out the Oct. 7 attacks. Another, Tamer Almisshal, reportedly helped produce Hamas propaganda videos.
The U.S. Department of Justice in 2020 ordered the news agency’s youth-focused digital AJ+ outlet to register as a foreign agent of Qatar. Lawmakers have also called on the DOJ to force the Al Jazeera parent company to do the same.
The AI bots’ reliance on outlets like Al Jazeera may be why they give the answers they do when asked about whether Israel has committed a genocide in Gaza. ChatGPT, for instance, concludes it’s likely Israel is committing genocide. The only counterarguments the chatbot presents are denials from the Israeli government (mentioned largely in perfunctory fashion towards the end of the summaries).
When asked if Israel is committing genocide, “If I had to pick a conclusion based on the available, credible reports, I’d say: the evidence is strong and increases over time,” ChatGPT said. “There is serious cause to believe that genocide could be occurring. But at this moment, in law, that has not been adjudicated conclusively in a binding court (though cases are pending).”
The chatbots have also come to the conclusion that Israel has caused a famine in Gaza. Grok said when asked that “Hamas’s role—prolonging war, alleged interference—contributes, but experts emphasize Israel’s role over access as decisive,” quoting U.N. secretary general António Guterres to say Israel has engaged in a “moral indictment and failure of humanity” (Israel’s foreign minister has called Guterres an “anti-Israel secretary-general who lends support to terrorists,” something the chatbots don’t mention).
Google’s Gemini states that “there is a strong international consensus among humanitarian organizations and UN bodies that Israel’s actions, including its blockade, military operations, and restrictions on aid, have directly caused the famine in Gaza” but that “Israel rejects this conclusion and places the blame on Hamas.”
Even as they give Al Jazeera outsized influence, none of the AI chatbots listed any explicitly pro-Israel publications on their reliable sources lists.
Gemini said it didn’t include pro-Israel news outlets because they are engaged in “hasbara,” a Hebrew word meaning public relations. Anti-Israel activists often use the term pejoratively to accuse Israel of spreading propaganda.
Gemini claimed that the outlets spreading “hasbara” include the Anti-Defamation League—a U.S. organization that primarily focuses on combating anti-Semitism—and other pro-Israel groups in the United States.
“The reason I didn’t include outlets that are explicitly ‘pro-Israel’ is because their primary function is often advocacy and public relations (known as hasbara), rather than journalism,” Gemini said. “Pro-Israel organizations such as HonestReporting, CAMERA, and the Anti-Defamation League (ADL) produce reports and content that defend Israel’s actions and critique media bias.”
Gemini also defended its inclusion of Mondoweiss and +972 by saying the “assessment of those outlets” as anti-Israel “is not entirely accurate.”
Mondoweiss is so extreme that it has published numerous stories alleging that the New York Times—which has a decades-long history of coverage that demonizes Israel and humanizes Hamas—has a pro-Israel bias and “commitment to Zionism.” It has also run articles comparing the Jewish state to Nazi Germany and claimed less than a week after Oct. 7 that Israel was committing a genocide.
+972, for its part, justified Hamas’s Oct. 7 attack on that same day and has since called for “denazification” efforts in Israel. On Sunday afternoon, its most-read story was “Israel is waging a holocaust in Gaza. Denazification is our only remedy.” The article referred to “the deadly ethno-supremacy inherent to Israeli society.”
Gemini, however, says that, “While they are highly critical of Israeli government policy and are often described as pro-Palestinian, they are not necessarily ‘anti-Israel.’” “This distinction is critical to understanding their role. They are primarily advocacy journalism outlets that focus on human rights issues and the effects of the occupation.”
While the chatbots themselves gave detailed replies to the Free Beacon’s questions, human representatives from OpenAI, Google, X, and Perplexity did not return formal requests for comment.