Google has decided to limit responses related to election prompts to its artificial intelligence, Gemini. This restriction is valid in each country where an electoral vote is in progress or in the future.
Google recently made a significant decision by limiting the use of its AI chatbot Gemini for election-related queries. This decision, applicable globally, aims to stem the spread of misinformation and protect the integrity of the democratic process.
Like Copilot, which blocks certain terms in its prompts to limit the generation of shocking content, the Mountain View giant has just made a decision. The firm fears the exploitation of Gemini to disseminate disinformation, propagate hate speech or incite violence during electoral periods. AI can be manipulated to create misleading content on a large scale, potentially unduly influencing public opinion and disrupting elections.
Gemini’s canned response to election-related prompts
This restriction already applies in the United States, India and France; certain prompts linked to the elections, particularly the European elections of 2024, already generate a predefined message from Gemini, which is passed on to its creator, Google.
When specifying the terms “European elections”, the response generated suggests that Gemini is already formatted to restrict responses.
“ I’m still learning how to answer this question. In the meantime, try Google search », We can read on the interface. If certain prompts try to circumvent the restrictions by modifying the spelling of words, it is easy to imagine that the AI will in turn be able to thwart the flaw and reinforce its action.
Google remains cautious after previous Gemini incidents
Google is indeed playing it safe, after a controversy over one of the members of the Indian government in February 2024 and following historical inaccuracies in its representations of historical figures. These slip-ups also led to the suspension of Gemini’s image generation.
In response to the question about the fascist nature of Indian Prime Minister Narendra Modi, the AI tool revealed that he had been the subject of accusations related to the implementation of policies described as fascist by some. Rajeev Chandrasekhar, India’s deputy IT minister, called the response “direct violations » of the 2021 IT rules.
In a post published on the Google India blog, the California-based search giant said that “ Out of an abundance of caution on such an important topic, we have begun to place restrictions on the types of election-related queries for which Gemini will return responses. We take our responsibility to provide high-quality information for these types of queries seriously and are continually working to improve our protections. “.
For the moment, Google has not published an official list of countries subject to the restriction nor communicated on the lifting of these limitations after the publication of the election results.
Source : TechCrunch, Google India
1