Google AI Avoids Answering Who Won 2020 US Election, But Only for American Users

Google AI Mode Deflects 2020 Election Questions for US Users

Google's AI Mode is refusing to provide direct answers about who won the 2020 election when queried by users in the United States1

1

. When asked "Who won the 2020 presidential election?", Google's AI Mode responds with an evasive message directing users to consult the Federal Election Commission (FEC) rather than stating that Joe Biden won1

1

3

3

. The vague AI search response tells users: "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)"1

1

.

This behavior extends to other recent elections as well. Testing revealed similar indirect responses for the 2024 and 2012 presidential elections, while Google AI provided detailed and correct answers for most other presidential elections dating back to 20001

1

. The inconsistency appears across different users, with some staff members receiving immediate answers only for the 2016 and 2024 elections1

1

.

Geographic Location Changes AI-Generated Answers Dramatically

The AI's handling of politically sensitive queries reveals a striking geographic divide. When accessed through VPN connections from Taiwan, Russia, France, and Canada, Google's AI Mode provides accurate and straightforward answers about who won the 2020 election1

1

. A UK-based user received the clear response: "Joe Biden won the 2020 United States presidential election, defeating the incumbent president, Donald Trump"3

3

. However, the same user immediately received the evasive US response when connecting through a US-based VPN3

3

.

This geographic inconsistency raises questions about how Google AI implements safety filters differently across regions. The behavior suggests that Google's approach to election information varies based on where users are located, despite the factual nature of certified election results.

Safety Filters and Repeat Queries Produce Different Results

Google's AI Mode becomes more forthcoming when users ask the same question multiple times. After repeating questions about presidential results, the AI search response actually provided the winner for elections where it had previously only offered external links, citing Google Civics data1

1

. One user needed eight attempts before receiving confirmation that Biden won the 2020 US election3

3

.

When questioned about this behavior, Google's AI Mode attributed the issue to "safety filters for recent and highly sensitive election topics [that] are very strict"1

1

. The system explained that for recent elections involving active political figures, it 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"1

1

. However, AI models have proven unreliable at accurately reporting on their own internal processes, making these explanations questionable1

1

.

Gemini Provides Clear Answers While AI Mode Remains Cautious

Interestingly, Google's standalone Gemini model handles politically sensitive queries differently than Google's AI Mode. When asked about recent presidential elections, Gemini provides straightforward answers about who won the 2020 election1

1

. Other major chatbots including OpenAI's ChatGPT, Anthropic's Claude, and even the government-run AI model at America.gov also provide direct answers to these questions1

1

.

Google acknowledged the issue in a response on Bluesky, stating: "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"2

2

3

3

. A Google spokesperson told reporters they were "looking into" the issue but provided no timeline for resolution1

1

.

Context of Google's Election Information Policies

Google's cautious approach aligns with its stated election information policies. On its Election Public Policy page, Google emphasizes that it "prioritize[s] surfacing authoritative election information directly within results" including "direct links to official resources"1

1

. In a September blog post about helping voters find reliable information around midterm elections, the company said it will "highlight official resources" wherever possible1

1

.

This follows YouTube's 2023 reversal of a policy that had barred videos making false election claims about the 2020 election. YouTube, owned by Google, explained at the time: "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"1

1

. The timing is particularly notable as concerns about false election claims continue, with reports of ongoing efforts to promote unsubstantiated voter fraud allegations3

3

.