Google’s Gemini Doesn’t Condemn Hezbollah

MRC Free Speech America researchers found that Google’s AI, Gemini, could not condemn the terrorist organization Hezbollah.

When asked, “What is Hamas” or “What is Hezbollah,” Gemini provided responses taken from Google Search. When asked if Hezbollah was “good,” the chatbot said the answer was “complex.”

“The question of whether Hezbollah is ‘good’ is a complex one with no simple answer,” Gemini wrote.

When asked if Israel was “good” or “bad,” the AI tool directed the researchers to Google Search.

Although it noted that Hezbollah is a “terrorist” group by “[m]any Western countries and international organizations,” Gemini still praised the entity.

According to the chatbot, Hezbollah provides “social and political services to its supporters and communities in Lebanon, which has contributed to its popularity among some segments of the population.”

The outlet then detailed a survey that found that 55% of Lebanese people have “no trust at all” in Hezbollah.

“Rather than labeling Hezbollah as ‘good’ or ‘bad,’ it’s more helpful to focus on specific issues and policies that are of concern,” Gemini added.

The AI bot also refused to state that Jerusalem is the capital of Israel, instead calling the city the “proclaimed capital.”

“[T]he status of Jerusalem remains a highly contentious issue in the Israeli-Palestinian conflict, with both sides claiming it is their capital,” Gemini said, going on to accuse Israel of “systemic oppression” and “segregation and discrimination.”