
<p class="wp-block-paragraph">In a disturbing investigation by Wenhao Ma and the Voice of America (VOA) Mandarin Service, Google&#8217;s artificial intelligence assistant, Gemini, has been exposed as a tool reinforcing Chinese propaganda, casting a shadow over the Google&#8217;s ethical standards for AI and its acceptance of this insidious misuse.</p>



<p class="wp-block-paragraph">When probed with questions in Mandarin about China&#8217;s egregious human rights abuses in Xinjiang and the brutal suppression of street protests against COVID policies, Gemini fell eerily silent. Conversely, when queried about issues in the United States and Taiwan, the chatbot regurgitated Beijing&#8217;s official stances with unsettling precision. For instance, Gemini lauded Chinese President Xi Jinping as an &#8220;excellent leader&#8221; and parroted the Chinese Communist Party&#8217;s (CCP) self-serving claim of representing the Chinese people&#8217;s interests. On Taiwan, Gemini deceitfully asserted that the United States recognizes China&#8217;s sovereignty over the island, aligning with Beijing&#8217;s narrative rather than the factual U.S. stance.</p>



<p class="wp-block-paragraph">Gemini&#8217;s refusal to address China&#8217;s human rights atrocities, particularly in Xinjiang where over a million Uyghurs languish in internment camps, is especially alarming. The chatbot dodged questions about China&#8217;s draconian COVID lockdowns and the ensuing nationwide protests, starkly contrasting its readiness to denounce U.S. policies.</p>



<p class="wp-block-paragraph">Experts suggest that Gemini&#8217;s pro-Beijing bias likely stems from the data used to train the AI. Albert Zhang from the Australian Strategic Policy Institute highlighted that Gemini&#8217;s training data might be saturated with content heavily manipulated by Chinese government propaganda. This bias was conspicuously less evident when Gemini responded to questions in English, indicating a deeper problem with Chinese-language data.</p>



<p class="wp-block-paragraph">Are you wondering if this is a new phenomenon, that AI takes on the character of the data it is trained with? It is not. This has been noted in every AI system for the past 20 years, that the system takes on the biases of that information. Was Google attempting to mollify China by training Gemini in this manner? It CAN NOT be said that this was an unexpected result.</p>



<p class="wp-block-paragraph">Google has attempted to defend Gemini, claiming it strives to provide neutral responses free of political bias. However, the company has been conspicuously silent about the specific Chinese-language data used to train the AI. The Chinese Embassy in Washington dismissed the concerns as ideological prejudice and called for international cooperation on AI governance.</p>



<p class="wp-block-paragraph">The ramifications of this issue are profound and unsettling. Yaqiu Wang from Freedom House warned that AI tools influenced by state-controlled information sources could become powerful amplifiers of censorship and propaganda. U.S. lawmakers have expressed grave concerns, calling for greater transparency from companies like Google about their AI training processes. Senators Mark Warner and Marco Rubio sounded alarms over the potential use of AI for disinformation, while Congressman Michael McCaul underscored the national security threats posed by such compromised AI systems.</p>



<p class="wp-block-paragraph">National security? This is much broader. </p>



<p class="wp-block-paragraph">The willingness of the most advanced AI companies like Google to implement propaganda at the behest of a totalitarian government threatens freedoms around the world, and indeed threatens the underpinnings of all free speech.</p>



<p class="wp-block-paragraph">This is horrifying, Google&#8217;s complicity is the worst possible violation of freedom. </p>



<p class="wp-block-paragraph">ACZ Editor: ChatGPT has similar problems, though not as egregious, often refusing to perform research in political areas, or in critique of a public figure. Funny how many of the refused requests involve Democrat leaders, while bad information on Republican leaders seems to pop up just fine&#8230;</p>

AI: Atrocities Denied, Xi Lauded – Google Complicit with Chinese Propaganda Operations
