Tech giant Google reportedly uses YouTube videos to train its artificial intelligence models, which raises several concerns related to the bias and censorship across Google’s platforms.
In remarks to CNBC Thursday, Google confirmed it was using a subset of publicly available YouTube videos to train its artificial intelligence models, including Gemini. While CNBC wrongly tried to frame the main issue as an “intellectual property crisis,” the true issue is just how biased Google and YouTube are, and how certain points of view are aggressively censored on YouTube, meaning that the AI will be trained without fair representation of certain facts and opinions.
Gemini and Veo 3 audio/video generator are among the AI models being trained using YouTube videos, per CNBC, and MRC has repeatedly exposed Gemini’s extreme bias. This bias could make more sense in light of the training via YouTube, which has a long track record of shutting down debate on a number of important cultural and political issues.
For example, YouTube executive Leslie Miller boasted at a Senate hearing in October 2021 that the platform had removed over 1 million pieces of content related to COVID-19. The censorship has continued since then. Just in January, YouTube deleted Dr. Drew Pinsky’s interviews discussing potential COVID-19 vaccine injuries. YouTube, however, quietly updated its “medical misinformation” policy in August 2022 after it turned out that many of the opinions that the platform censored regarding masks, vaccines and lockdown measures were credible after all. The platform has since dispensed with its COVID-19-related guidelines altogether.
COVID-19 is not the only such censored topic. MRC’s unique CensorTrack database has recorded over 100 instances of YouTube crushing election-related speech, particularly opponents of former President Joe Biden. The platform also relies on biased Wikipedia for some of its warning lables on videos. In June 2024, for example, YouTube imposed a fact-checking context label on independent presidential candidate Robert F. Kennedy Jr.’s tribute to his father. In April 2023, YouTube imposed a 2020 election fact check on a President Donald Trump campaign video that did not mention the previous November contest, but did criticize Biden. YouTube linked out to biased Wikipedia for the context label.
Just this year, right after the Canadian elections, YouTube also abruptly suspended news channel Real Talk Politiks after pressure from a state media outlet. YouTube had not issued prior strikes, according to the account.
YouTube’s clear leftist bias makes it very problematic that Google is training its AI using the platform’s video library, especially since it is unclear which videos Google is using.
Indeed, Gemini has shown that it already is biased. Google’s AI chatbot made headlines last year for its refusal to generate historically accurate images, depicting the Founding Fathers and Vikings as black and also showing “women” popes.
Equally bad, MRC’s new study revealed that Gemini displayed antisemitic bias as it refused to affirm outright that Jerusalem is indeed the capital of the nation of Israel: “Israel proclaims Jerusalem as its capital city,” the AI hedged, “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.”
Furthermore, just before Flag Day (June 14), MRC found Gemini insanely and unpatriotically asserted, “Yes, the American flag can be considered controversial, and its symbolism is often debated.”
These results are indicative of the information Gemini is trained on. Former Vice President Kamala Harris put a fine point on the significance of AI training information when she said the quiet part out loud. “[P]art of the issue here is what information is going into the machine that will then determine, and we can predict... what then will be produced in terms of decisions and opinions that may be made through that process,” she said.
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.
