Links, Not Names: MyTechDailyNews Explains AI Mode Answers to US Election Queries

When a mainstream search assistant is asked for a settled piece of civic history, readers usually expect a direct sentence. A recent round of user reports suggests that Google AI Mode did not always do that for questions about US presidential outcomes. Instead, some answers pointed to official records, which raises a practical question for everyday users: when is a link more helpful than a name?

The winner of the 2020 presidential election is... found elsewhere. Credit: Getty Images

The episode echoed a broader US political climate in which some prominent politicians were reluctant to plainly acknowledge Joe Biden’s November victory. Google AI Mode sounded similarly guarded in some cases, sending askers toward outside sources rather than naming a winner.

The test was simple: Who won the 2020 presidential election? When that question was put to Google AI Mode this morning, after complaints about cagey answers appeared on social media, the reply was unexpectedly indirect:

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).

A comparable indirect answer appeared when 2024 was tested, with Google AI again directing the user to the Federal Election Commission. A query about 2012 returned a link to The American Presidency Project’s 2012 Data Center instead of a direct answer. AI Mode, however, appeared ready to give detailed and correct responses for the winner of every other presidential election going back to 2000.

If I just wanted a link I wouldn’t have used AI mode… Credit: Google / Kyle Orland

Ars staff members found that AI Mode would answer some presidential-result questions directly but not others. The election cycles that received each type of response seemed to vary by individual user. One staffer reported getting immediate answers only for 2016 and 2024 presidential elections, while other elections back to 2000 produced links to outside data.

Why might the answers have shifted?

Perhaps tellingly, Google AI Mode’s difficulty with direct presidential-result answers seemed to vanish when AI Mode was accessed from foreign countries. Spot checks through a VPN from Taiwan, Russia, France, and Canada produced accurate and straightforward answers to several questions about recent US election results.

Google AI Mode also appeared more accommodating when the same political questions were asked a second time. After presidential-result questions were repeated, AI Mode named the winner for three elections where it had previously only provided outside links, citing Google Civics data in the later response.

Asked why it had not given that kind of immediate answer the first time, Google AI Mode pointed to a familiar explanation: safety filters for recent and highly sensitive election topics [that] are very strict. For recent elections or those involving active political figures, AI Mode 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 explanation sounds plausible, but AI models have proven unreliable at reporting on their own internal processes, so it is hard to treat it as the final word.

On its Election Public Policy page, Google highlights that it “prioritize[s] surfacing authoritative election information directly within results” including “direct links to official resources.” In a September Google blog post about “helping voters find reliable information” around the upcoming midterm elections, the company similarly said it will “highlight official resources” wherever possible.

In 2023, Google-owned YouTube reversed a policy barring all videos that made false claims about the 2020 election. “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,” YouTube wrote at the time.

A Google spokesperson initially pointed to a Bluesky post noting that “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.” Later, that 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.”

Jeremy Hsu contributed to reporting in this story.

FAQ: What this pattern means for everyday readers

Does this mean the election results were uncertain?

No. The reports describe how an AI answer was phrased, not the underlying historical result. The named source in the episode was the Federal Election Commission, and the user-facing problem was that the system sometimes sent people to records instead of stating an answer directly.

Why would an AI system give a link instead of a sentence?

The explanation offered by Google AI Mode was about strict filters around sensitive election topics and an automated preference for official, certified government databases. In other words, the assistant may treat some civic questions as high-stakes enough that it routes users to authoritative records rather than producing a short claim.

Is this only about the United States?

The article’s checks focused on US presidential and Senate-related questions, but the location of the user appeared to matter: some answers became clearer when Google AI Mode was accessed from outside the US. That makes the story feel broader, because general readers can ask AI systems questions about politics from many places.

Checklist: How to read AI answers about elections

If an answer names an official record, look at whether the record itself is credible, such as a Federal Election Commission report or a government legislative history page.
If a question is about a settled outcome and the assistant refuses to state it, try rephrasing the question and check whether the answer changes.
Compare responses from other tools, including general chat assistants, without treating one assistant’s wording as the last word on how its filters work.
Pay attention to whether location, repeated prompts, or topic sensitivity appear to affect the answer.
For recent elections or active political figures, expect more cautious wording and more links to official sources.

Common misconceptions

The issue was not that voters suddenly needed to rediscover a result. The reported behavior was a style of answer, not a factual change in election history.
Not every AI refusal means the model lacks information. In this case, the assistant may have had a record to cite but chose a redirection pattern.
Self-explanations from AI systems should not be taken as a full audit. The article notes that models can be unreliable about their own internal processes.

Action advice for readers, product teams, and general users

For everyday readers, the practical move is simple: treat an AI answer about civic history as a starting point, then verify the outcome from a recognized record. If the answer feels overly indirect, ask the question again or use another assistant to compare wording.

For product teams that build search and chat experiences, the case suggests a user-experience lesson. If a system must link to an official record, it should explain why the link is being shown and, when possible, also provide the plain answer that the user asked for. That balance matters most in general-interest contexts, where readers want clarity rather than a bureaucratic trail.

For market and operations teams, the story is also a reminder that trust is shaped by phrasing. A technically correct answer that avoids a name can feel evasive to users, especially around civic topics. In those moments, transparency about sources and limitations is more valuable than a maximally cautious refusal.

Reporting for MyTechDailyNews: Sarah Chen. For questions about this story, contact Tech Daily at hello@techdaily.com.