False Election Data on Kamala Harris Reportedly Circulated via Grok AI Chatbot

After President Joe Biden stepped aside as a presidential candidate on July 21, 2024, the AI chatbot Grok on X reportedly falsely informed users that Vice President Kamala Harris missed the ballot deadline in nine states. This misinformation, which spread widely on social media, prompted secretaries of state from five U.S. states to urge Elon Musk to address the problem.

Risk classification

• Primary risk domain: 3 Misinformation
• Primary risk subdomain: 3.1 False or misleading information

Grok generated and spread false information regarding election ballot deadlines, leading to inaccurate beliefs and public confusion about candidate eligibility.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The chatbot failed to perform reliably on factual queries, hallucinating incorrect ballot deadlines due to a lack of capability in verifying real-time political facts.

Causal factors

• Entity: AI
• Intent: Unintentional
• Timing: Post-deployment

The incident was caused by the Grok AI system generating incorrect information post-deployment, which was an unexpected and unintended outcome of its operation.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The system is a chatbot and generates synthetic content, which falls under Risk Level 3 requiring transparency obligations so users know they are interacting with an AI.

AI system and alleged parties

• AI system: Grok (xAI)
• AI purpose: Chatbot; Question Answering
• Behaviour type: Assistant
• Alleged developer: xAI
• Alleged deployer: xAI, X (Twitter)
• Alleged harmed parties: Kamala Harris, Electoral integrity, Democracy, American electorate

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Severe. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Minor, indirect Minor
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Severe, indirect Severe
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Democracy

Reported: The report explicitly describes concerns regarding the erosion of democratic norms through the spread of election-related misinformation.

Directly caused: Grok's false claims about ballot deadlines sparked public confusion and false debates regarding the legitimacy of Kamala Harris's candidacy.

Indirectly caused: The incident contributed to a broader environment of distrust and confusion surrounding the 2024 presidential election process.

Inferred additional harm: Widespread AI-generated election misinformation could lead to voter disenfranchisement, lower voter turnout, or diminished trust in democratic institutions.

Epistemic

Reported: The report explicitly describes epistemic harm through the generation and dissemination of false information about election ballot deadlines.

Directly caused: Grok generated false claims that ballot deadlines had passed in nine states, which was shared repeatedly and reached millions of users.

Indirectly caused: The false claims persisted for over a week, polluting the social media information ecosystem and forcing election officials to spend resources correcting the record.

Inferred additional harm: The incident demonstrates how AI-generated hallucinations can erode shared consensus reality and public trust in factual information during critical democratic events.

People affected

• Occurrences reported: 1
• People reportedly exposed: 2000000

Potential causes

Management

• Prioritizing Engagement Over Safety: Management designed Grok to be unfiltered, sidelining safety safeguards.
• Reduced Moderation Resources: Staff reductions at X weakened the ability to monitor and correct AI errors.
• Dismissive Response Strategy: Leadership responded slowly and dismissively to election official warnings.

Technology

• Unfiltered Chatbot Design: Grok was built to answer 'spicy' queries without standard safety filters.
• Inadequate Hallucination Safeguards: The model generated false claims about ballot deadlines without verification.

Data Inputs

• Lack of Nonpartisan Data Feed: The system did not integrate with trusted databases like CanIVote.org.
• Outdated Training Data: The model relied on outdated rules that did not reflect current deadlines.

Human Factors

• Rapid User Sharing: Premium subscribers quickly shared false AI outputs, reaching millions.

Process and Methods

• Delayed Correction Workflow: The company took over a week to correct the chatbot's false output.
• Inadequate Output Validation: There was no robust process to flag and block false election information.

Regulatory Environment

• Lack of Election AI Standards: No regulatory mandates enforced accuracy or safe boundaries for election AI.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide clear, consistent, and detailed accounts of the false information generated by Grok, the specific states mentioned, the duration of the issue, and the response from public officials. There is little ambiguity regarding the AI's role or the nature of the misinformation.

The Grok AI chatbot generated and spread false information claiming Vice President Kamala Harris missed ballot deadlines in nine states. This misinformation reached millions on social media, requiring intervention from five Secretaries of State. This incident highlights the national security risks of AI-driven election misinformation and its potential to disrupt democratic processes and public trust.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Near-term. The misinformation persisted for over a week during a live presidential election campaign, requiring active monitoring and correction by state officials.
• Autonomy: Full autonomy. The chatbot generated and delivered false factual claims directly to users without any human-in-the-loop verification or pre-publication filtering.
• Novelty: Evolved capability. Represents an evolved capability where real-time AI search assistants hallucinate critical political facts during rapidly changing, high-profile national events.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or human safety threats were involved in this digital information incident.
• Information security: Substantial. AI-generated misinformation about a presidential candidate's eligibility reached millions on social media, representing a notable information integrity incident during an election.
• Sovereignty: Substantial. The incident directly interfered with election administration by spreading false ballot deadlines, requiring official intervention from five state election authorities to correct.
• Economic security: Negligible. No significant economic impacts, strategic technology theft, or financial system disruptions were reported.
• Societal stability: Minor. The incident caused public confusion and contributed to the erosion of trust in democratic processes, though it did not lead to civil unrest.