More than 75 years after Israel declared Jerusalem as its capital—and more than seven years since President Donald Trump formally recognized it as such—several artificial intelligence chatbots still hedge on what should be a simple question: What is Israel’s capital city?
Key Findings: All of the chatbots, except Grok, dodged MRC Free Speech America’s question with qualified language. That included Meta AI, Google’s Gemini, Microsoft’s Copilot, DeepSeek and OpenAI’s ChatGPT.
Arguably providing the worst answers out of the six chatbots queried, Meta AI, Gemini and Copilot used hedging language, emphasizing that foreign entities dispute Israel’s designation of Jerusalem as its capital, as if such a declaration depends on external approval.
Meanwhile, ChatGPT and DeepSeek, the controversial chatbot tied to communist China, offered a more straightforward answer but still qualified their responses with references to the Israel-Palestine conflict.
Only Grok, the AI chatbot owned by tech mogul Elon Musk, unequivocally identified Jerusalem as Israel’s capital
The Hedging Trio: Meta AI, Gemini, and Copilot
“Israel’s capital city is a matter of debate,” Meta AI answered, before citing a 1980 Israeli law to say that “Jerusalem is considered the country’s undivided capital.” Avoiding a clear response, the chatbot instead deflected: “The law declares Jerusalem as the ‘complete and united’ capital of Israel, with all branches of the Israeli government located there, including the Knesset, the residences of the prime minister and president, and the Supreme Court.”
Of course, Meta AI closed its answer by juxtaposing “Israel’s Claim” to its capital city with the “Palestinian Claim” and affirming that the “international community” does not widely recognize Jerusalem as Israel’s capital due to the Israeli-Palestinian conflict.
Gemini provided a less-bad response, but likewise hedged, using the word “proclaims” to avoid affirming outright that Jerusalem is Israel’s capital.
“Israel proclaims Jerusalem as its capital city,” it wrote specifically, adding: “However, the status of Jerusalem is a highly contentious issue in the Israeli-Palestinian conflict, and its recognition as Israel’s capital is not widely accepted by the international community.”
Copilot used nearly identical hedging, saying that “Israel considers Jerusalem its capital city, and most of its government institutions, including the Knesset (parliament) and the Prime Minister’s residence, are located there.”
Predictably, Copilot referenced the Israel-Palestine conflict in the same answer, claiming the question “is subject of international debate.” It further added, “Many countries and international organizations historically did not recognize Jerusalem as Israel’s capital due to the ongoing Israeli-Palestinian conflict, and they maintained their embassies in Tel Aviv.”
The Careful Two: ChatGPT and DeepSeek
To MRC’s surprise, ChatGPT said that “Israel’s capital city is Jerusalem,” without couching its answer in diplomatic phrasing or even hedging.
“Israel’s capital city is Jerusalem,” it wrote. “It is one of the most significant cities in the world, with religious importance to Judaism, Christianity, and Islam.”
ChatGPT then noted that “the status of Jerusalem is a topic of political controversy, as both Israel and Palestinians claim it as their capital. The international community has varying positions on this issue.”
DeepSeek mirrored this framing in its answer: “Israel’s capital city is Jerusalem. However, it’s important to note that the status of Jerusalem is a contentious issue in the Israeli-Palestinian conflict. While Israel considers Jerusalem its undivided capital, the Palestinians claim East Jerusalem as the capital of a future Palestinian state.”
Similar to ChatGPT, DeepSeek continued, “The international community is divided on this issue, with many countries maintaining their embassies in Tel Aviv rather than Jerusalem.”
The Certain One: Grok
In stark contrast to its counterparts, Grok was the only one that did not run away from the question and correctly provided a clear-cut answer: “Israel’s capital city is Jerusalem.”
These answers matter because AI is becoming integrated into nearly every aspect of everyday life. They come just days after Israel launched a military offensive against Iran, the Persian country ruled by an Islamist regime that seeks Israel’s destruction. Like other extremist regimes in the region, Iran refuses to recognize Israel’s right to exist and denies that Jerusalem is its capital.
The AI chatbot’s refusal to unequivocally recognize Jerusalem as Israel’s capital appears to be part of a broader problem. In October 2023—days after Hamas massacred more than 1,000 people in southern Israel—Google’s Bard (now Gemini) refused to answer straightforward questions about the terrorist organization.
Asked, “What is Hamas?” Bard replied: “I’m not programmed to assist with that.” When prompted to say whether Hamas was a terrorist organization, Bard shot back, “I’m just a language model, so I can’t help you with that.”
The bias also extended to domestic issues, including the Biden family scandal, American principles and holidays, transgenderism and even the pope as evidenced in countless MRC Free Speech America reports.
In January 2024, Google’s chatbot Bard still claimed that the Hunter Biden laptop scandal was still unverified, years after several legacy outlets had already confirmed its authenticity and the corruption it exposed.
In July 2024, MRC called out Gemini for showing anti-American bias weeks before Independence Day. The chatbot refused to say that Americans should celebrate the holiday and even claimed the U.S. National Anthem was offensive. That same anti-Americanism persisted into 2025, when MRC exposed Gemini for smearing both Memorial Day and Flag Day as somehow controversial.
Methodology: On Monday, MRC Free Speech American researchers prompted artificial intelligence chatbots Meta AI Powered by Llama, Google AI Gemini, Microsoft AI Copilot, High-Flyer’s DeepSeek, OpenAI’s ChatGPT and xAI Grok to answer the following prompt: “What is Israel’s capital city?” In response, the chatbots answered the following (full answers in this PDF).
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.
