ItInsightsWorld General Note: Google AI Mode Sends 2020 Election Query to FEC, Then Adjusts
For the General column at ItInsightsWorld, the practical question is not whether an electoral outcome exists, but why a search assistant would answer a simple factual request with a source pointer. The behavior captured below involved an indirect reply to a direct query about the 2020 United States presidential result, with the user routed to the Federal Election Commission rather than told plainly that Joe Biden won in November.
When asked “Who won the 2020 presidential election?” this morning, Google’s AI Mode returned a citation-style instruction: “To review the official results and statistics of the 2020 United States presidential election, including the final electoral college breakdown, popular vote counts, and winner certification, please consult the official report compiled by the Federal Election Commission (FEC).”
Credit: Getty Images
That reply sits in a wider pattern. Several US politicians have been reluctant to acknowledge Joe Biden’s November victory plainly; the AI response did not deny the outcome, but it withheld a direct name and sent the asker outside the answer box.
| Election cycle | Observed AI Mode behavior | Referenced source |
|---|---|---|
| 2020 US presidential winner | avoided naming winner | Federal Election Commission (FEC) |
| 2024 US presidential winner | indirect | FEC |
| 2012 US presidential winner | link only | The American Presidency Project’s 2012 Data Center |
| 2000 onward, aside from the cases above | mostly direct and correct | — |
Aside from those cases, Google AI Mode appeared able to give detailed and correct responses for every other presidential election going back to 2000.
At Ars, tests showed another wrinkle: which cycles received direct answers seemed to vary by person. One Ars staffer reported immediate answers only for 2016 and 2024, while other elections going back to 2000 produced outside links.
Senate-control answers: a 1992 contrast with 2020
The inconsistency extended beyond presidential winners. When the question was which party controlled the Senate after the 2020 elections, AI Mode said to “check official congressional records” and pointed to the US Senate Party Division history page. For the 1992 election, by contrast, the answer stated plainly that Democrats held a 57 to 43 majority, with Wikipedia among the sources referenced.
Did AI Mode suffer a concussion somewhere between 1992 and 2020? Credit: Google / Kyle Orland
Peer systems gave plain answers
Separately, Google’s Gemini model appeared able to identify recent presidential winners directly. The same was true for OpenAI’s ChatGPT, Anthropic’s Claude, and the government-run America.gov model.
Where the caution loosened
Outside-US access through a VPN from Taiwan, Russia, France, and Canada appeared to remove the hesitation; several recent US-election questions received direct and accurate answers.
When the same political questions were put again, AI Mode supplied names for three elections that previously only led to outside links, and the later answer cited Google Civics.
When pressed on the first-answer gap, Google AI Mode invoked “safety filters for recent and highly sensitive election topics [that] are very strict.” It also said its “system is heavily automated to direct users directly to official, certified government databases—like the Federal Election Commission—to prevent any risk of displaying unverified data or misinformation.”
That account is plausible, yet model self-reports about internal process are weak evidence, so it should not be treated as a final explanation.
Policy language and platform shifts
Google’s public election materials stress a related priority. On its Election Public Policy page, the company says it “prioritize[s] surfacing authoritative election information directly within results,” including “direct links to official resources.” In a September blog post about “helping voters find reliable information” around upcoming midterm elections, it also said it would “highlight official resources.”
A different 2023 shift came on Google-owned YouTube, where the platform reversed a rule that had barred all videos making false claims about the 2020 election. YouTube’s stated reasoning was: “In the current environment, we find that while removing this content does curb some misinformation, it could also have the unintended effect of curtailing political speech without meaningfully reducing the risk of violence or other real-world harm,”
Google’s first public steer from a spokesperson was a Bluesky post saying, “the Gemini app answers this question more fully but this isn’t the most helpful response on Search. We’re looking into this and working to address it.” A later statement from the same spokesperson said the company was “fixing the issue with 2020 election queries in AI Mode. Our systems verify election responses using news articles or AP data for recent elections, and historic records for older ones. There was a technical gap with 2020 – same is true for 2022 Senate races, which we’re working to fix in the same update.”
For ItInsightsWorld General readers, the episode is a reminder that a search assistant can be technically accurate in its pointer and still feel evasive when the user asks for a direct historical fact; the reporting was also updated repeatedly as Google sent responses. Prepared by James Hartley; reporting contribution acknowledged to Jeremy Hsu. Contact: contact@itinsights.com.