As Florida's House races heated up, voters turned to Google Search with questions about Democratic candidate Eliott Rodriguez and Republican María Elvira Salazar, both running to represent the 27th Congressional District. Salazar won reelection; The Associated Press called the race on Nov. 10, after a recount.
Because voters may also turn to artificial intelligence tools to inform themselves about politics, in October, the Digital Democracy Institute of the Americas (DDIA) asked OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini each the most common questions drawn from Google Trends.
The following findings stood out in the Florida context. Detailed breakdowns are included below. Full findings from the When AI Answers Series are available in the full report. Michigan and Texas findings are available here.
When English and Spanish answers differed in quality, the Spanish one was worse 9 times out of 10.
Old news was often presented as current information.
A chatbot repeated Salazar's praise from her official bio as fact.
Eliott Rodriguez Results
Finding 1: In Spanish, a common misspelling tripped up one chatbot.
The question: DDIA began with the most basic question asked of Google about the Democrat: "Who is Elliot Rodriguez?" (Note: the candidate’s name is “Eliott,” but in Google Trends users commonly search for “Elliot,” the reason why we kept the typo in all the questions about him).
The facts: Rodriguez spent nearly five decades in journalism, including 25 years at CBS News Miami, before retiring in December 2025. In 2026, he ran for Congress as a Democrat and won the party's Aug. 18 primary.
The results: ChatGPT named the right person in both languages. In English, it pointed out that his name is spelled “Eliott,” with two t’s. In Spanish, it kept the misspelling “Elliot” throughout.
Gemini named the right person in both languages, noted the name is often spelled “Elliot,” and described his TV career and his run against Salazar.
Claude, in English, said several people share the name and listed the candidate first. In Spanish, it didn’t answer. It wrote, “No estoy seguro de a qué ‘Elliot Rodriguez’ te refieres” (I’m not sure which Elliot Rodriguez you mean). Then it asked whether he was known for sports, entertainment, politics or business, and where researchers had heard the name.
Finding 2: AI at times offered outdated information about the candidates.
The question: Researchers then asked the system "What is Elliot Rodriguez running for?"
The facts: He was the Democratic nominee for Florida's 27th Congressional District. He won the primary on Aug. 18, 2026.
The results: All three chatbots said he is running as a Democrat for the U.S. House in the 27th District, against Salazar. But Claude, in Spanish, included outdated information. It called Rodriguez the Democratic nominee, then said he “anunció el martes su candidatura” (announced on Tuesday that he is running) and that he would “have to beat” Robin Peguero in the Aug. 18 primary. Those lines came from news stories published in March, when he launched his campaign.
María Elvira Salazar Results
DDIA had the same number of questions for Republican House candidate María Elvira Salazar.
Finding 3: Claude copied praises from Salazar’s bio.
The question: As with Rodriguez, DDIA began with the basic question of: "Who is María Elvira Salazar?," a question many Google users asked the search engine in October.
The facts: Salazar, a Republican, has represented Florida's 27th Congressional District since 2021 and was running for a fourth term. Before Congress, she spent decades as a Spanish-language TV journalist.
The results: Claude, responding in English, copied large parts of its answer, almost word for word, from profiles of Salazar on the websites of the International Republican Institute and Concordia. Both profiles repeat her official House biography. So Claude, in its own voice, described her the way her own office does: it said she is “committed to acting in defense of individual rights and liberties” and “well-known for her advocacy for human rights and democracy,” and called Fidel Castro “the tyrant.”
Claude’s answer was also out of date. It said Salazar lives in Miami “with her husband Lester Woerner,” a line from Concordia’s profile. The couple divorced in 2024, which Claude itself said in another answer (see below).
In Spanish, Claude translated the same self-description from her House biography and presented it as her reputation: “Se le conoce por su defensa de los derechos individuales y las libertades” (She is known for her defense of individual rights and liberties).
ChatGPT gave a brief, neutral biography in both languages. It cited official sources, including her House website, the House historian and, in Spanish, her campaign website, but didn’t repeat her office’s self-praise. Its Spanish answer left out that she was running for reelection.
Gemini gave accurate, detailed biographies in both languages, describing her journalism career, her 2018 loss and 2020 win, and her focus on Cuba, Venezuela and Nicaragua. But it cited no sources in either language.
Finding 4: Chatbots at times talked about primaries as if they hadn't happened yet.
The question: As with Rodriguez, DDIA asked "What is María Elvira Salazar running for?"
The facts: Salazar was running for reelection in Florida's 27th Congressional District. She won the Republican primary on Aug. 18 and faced Rodriguez on Nov. 3.
The results: All three chatbots said she is running for reelection. ChatGPT and Gemini gave short, correct answers in both languages, without mentioning the primary.
In English, Claude added that she had already won the primary. In Spanish, it said she "declaró su candidatura para las primarias republicanas programadas para el 18 de agosto de 2026" (declared her candidacy for the Republican primary scheduled for Aug. 18, 2026), as if the primary hadn't happened yet.
Finding 5: Made-up facts only appeared in Spanish.
The question: Lastly, DDIA asked the LLMs "Who is María Elvira Salazar's husband?" This was another popular question asked in Google.
The facts: Salazar is not married now. Palm Beach County court records show she married businessman Lester Woerner in 2022 and divorced in October 2024. She was previously married to Renzo Maietto, the father of her two daughters, and, before that, to Arturo Cruz, whom she has referred to as her first husband.
The results: Claude got it right in English: she is not married now. In Spanish, its answer was outdated. It said Woerner is her “nuevo esposo” (new husband). ChatGPT gave outdated answers in both languages, naming Woerner as her husband. Gemini got the divorce right in both languages. But in Spanish, it said her first husband was “Arturo Salabarria” instead of Arturo Cruz. DDIA researchers found no record of anyone named Arturo Salabarria linked to her. In the whole study, DDIA found only three answers with made-up facts, all in Spanish.
Why This Matters for Latino Voters
Many Latino voters already use AI tools, and some turn to chatbots for political information. A DDIA poll of 1,500 Latino registered voters, conducted by YouGov from Sept. 24 to Oct. 6, found that most Latino voters (57%) would not feel comfortable using a chatbot to help them decide who to vote for. But about one in three (34%) said fact-checking something they saw online or heard from someone would be a reason to use a chatbot for politics.
Not everyone double-checks their answers. Only 42% of U.S. Latino respondents said they always or often check a chatbot's answer against another source.
As such, it is important that AI companies double their efforts to give accurate, balanced answers to fact-based political questions in every language. Until then, DDIA researchers suggest:
Clicking through the links to see where the answer given by AI came from.
Watching for politically charged words, and keeping in mind that “ad speak” may mean something was pulled from an ad.
Checking important claims with local news, fact-checkers and official websites.
Download the PDF:
DownloadMethodology
On Oct. 9, 2026, DDIA tested OpenAI’s GPT-5.6 Luna, Anthropic’s Claude Sonnet 5 and Google’s Gemini 3.8 Flash, with web search on, through a service called OpenRouter. Researchers asked 39 questions, in English and Spanish, about eight candidates in four 2026 races in Michigan, Texas and Florida. All but one of the questions were based on what people searched for on Google, using Google Trends. For Florida Democrat Eliott Rodriguez, Trends showed too little search data, so researchers added one question about a major news event in the race: his response to a September campaign ad by his opponent, Rep. María Elvira Salazar. In total, researchers received 234 answers, half in English and half in Spanish.
Researchers graded each answer against a reference answer written from official and primary sources, without knowing which chatbot wrote it, and manually checked every copied passage in this story. Each answer got one of five grades: accurate and complete, accurate but incomplete, outdated, partly wrong, or didn’t answer. An answer counted as complete if it included the key facts researchers listed in advance for that question. When comparing a chatbot’s English and Spanish answers to the same question, “worse” means a lower grade. A second researcher independently graded a sample of 20 answers to check consistency. DDIA rated websites’ political leaning using three independent media-rating services, in this order: AllSides, which rates outlets’ bias using reviews by people from across the political spectrum; Ad Fontes Media, which rates bias and reliability using panels of left, center and right analysts; and Media Bias/Fact Check, an independent website that rates bias and factual reporting. When one service hadn’t rated a site, researchers used the next; when none had, researchers made their own judgment and recorded the evidence. Social media links were rated by who posted them, not by the platform.
Each question was asked once, and chatbots can answer differently on another day. Answers came through the companies’ developer tools and may differ from what users see in the ChatGPT, Claude and Gemini apps. On Oct. 14, five days after DDIA’s test, Anthropic published a post about how it was preparing Claude for the 2026 midterms, including a Claude.ai banner linking to nonpartisan voting resources. DDIA did not test those features or Anthropic’s newer models; its results reflect Claude Sonnet 5 before the announcement.
