Google confirms global restrictions on Gemini election queries

If you want to ask Google’s AI chatbot a question about the upcoming election, you’ll have to ask in a country where the election isn’t happening. TechCrunch has learned that the search giant has begun limiting election-related queries made by Gemini, such as any Global markets are holding elections.

The search giant confirmed to TechCrunch that it is starting to roll out restrictions on Gemini to limit answers to election-related queries globally. TechCrunch has learned that the update is already live in the United States and is starting to roll out in India and all major countries where elections will be held in the coming months.

The shift highlights not only the role that generative artificial intelligence has played and could play in the election process, but also Google’s concerns about how the service might be weaponized and generate inaccurate or misleading responses.

The shift comes just days after India issued a recommendation requiring technology companies to obtain government permission before rolling out new artificial intelligence models.

Queries about political parties, candidates, and politicians now return default messages.

When asked a question about a specific party or candidate, Gemini displays a message: “I’m still learning how to answer this question. In the meantime, try a Google search.” In some cases, the tool can still be exploited. TechCrunch found that the AI ​​tool did show answers when passed queries with misspellings. In a game of rapid engineering whack-a-mole, reactions will likely continue to adapt.

Google Gemini restricts answers to election-related queries (left), but not when the question contains typos (right) Image Source: Jagmeet Singh/TechCrunch

“Out of an abundance of caution on such an important topic, we have begun limiting the types of election-related queries to which Gemini will return responses. We take our responsibility to provide high-quality information for such queries seriously and are continually working to improve our protections measures,” Google said in a blog post related to India’s move earlier today.

The update comes ahead of the announcement of India’s general election. Earlier this month, New Delhi issued an advisory to technology companies restricting them from releasing new artificial intelligence models in the country without government approval. After facing a backlash from global venture capital and artificial intelligence startup founders, the Indian government clarified that the restrictions were aimed at “important” technology companies and not startups.

The advice comes after a mini-scandal involving Gemini and the Indian government last month. The artificial intelligence tool responded to a question about whether Indian Prime Minister Narendra Modi is a fascist, saying that Modi has been accused of implementing policies that some have characterized as fascist. Rajeev Chandrasekhar, Deputy Minister of Information Technology, India Call it a “direct violation” IT Rules 2021.

Last month, Google also suspended Gemini’s ability to generate images of people because it displayed historical errors. The company later said it would soon re-release an improved version to address these issues.

It’s unclear whether Google will unlock Gemini after the election later this year in response to election-related questions. We’ve reached out to Google about this story and will update this story when the company responds. We’re still waiting for the full list of countries currently being updated and we’ll update the post as we learn more and post as is.