AI Chatbots Reportedly Inaccurately Conveyed Real-Time Political News

Over a week of back-to-back, significant breaking political news stories, including the Trump rally shooting and Biden’s campaign withdrawal, AI chatbots reportedly failed to provide accurate real-time updates. Most chatbots gave incorrect or outdated information, demonstrating their current limitations in handling fast-paced news. These incidents suggest the continuing need for improved AI capabilities and caution in their deployment for real-time news dissemination.

Risk classification

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

The AI chatbots generated false or misleading information regarding major breaking news events, such as denying that an assassination attempt on Donald Trump had occurred.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The chatbots demonstrated a lack of capability and robustness in handling real-time, fast-paced breaking news events reliably.

Causal factors

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

The incident was caused by deployed AI chatbots unintentionally generating incorrect, outdated, or evasive responses to real-time news queries.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Risk Level 3: Limited Risk. The systems described are 'AI chatbots' which are subject to transparency obligations under the Act to ensure users are aware they are interacting with an AI.

AI system and alleged parties

• AI system: ChatGPT, Copilot, Gemini, Meta AI, Perplexity (Google, Meta, Microsoft, OpenAI, Perplexity AI)
• AI purpose: Chatbot; Question Answering
• Behaviour type: Assistant
• Alleged developer: Perplexity, OpenAI, Meta, Google
• Alleged deployer: Perplexity, OpenAI, Meta, Google
• Alleged harmed parties: Journalism, General public, Chatbot users

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. 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 Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Minor
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Epistemic

Reported: Yes, the report explicitly describes chatbots generating false and misleading information about major breaking news events.

Directly caused: ChatGPT falsely claimed that rumors of an assassination attempt on Donald Trump were misinformation, and Perplexity stated there were no reports of anyone being shot.

Indirectly caused: Chatbots listed Joe Biden as still running for president after he had withdrawn, potentially misleading users who relied on them for real-time updates.

Inferred additional harm: It is highly likely that many of the millions of users interacting with these default search tools were misinformed about these critical breaking news events before corrections were made.

People affected

• Occurrences reported: 1

Potential causes

Management

• Prioritizing Caution Over Utility: Companies block election queries out of an abundance of caution.
• Premature Search Integration: Baking unoptimized chatbots into default search features across platforms.

Technology

• Post-hoc Citation Generation: Sourcing is added after the fact, causing reference and accuracy issues.
• Delayed Update Capabilities: Chatbots are not real-time products and require time to update databases.
• Optimization for Engagement: Systems are optimized for conversational engagement rather than factual truth.

Data Inputs

• Information Vacuum: Lack of immediate credible data right after breaking news events occurs.
• Unreliable Real-time Pulling: Pulling from early, unverified reports leads to displaying false claims.

Human Factors

• User Reliance on AI for News: Users turn to conversational AI for instant answers during breaking news.

Process and Methods

• Inadequate Real-time Verification: No process exists to quickly verify and update facts during breaking events.
• Aggressive Query Blocking: Redirecting political queries to search engines instead of answering them.
• Sudden Deletion of Correct Info: Meta AI deleted correct answers to replace them with generic messages.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear, direct testing results of multiple specific AI chatbots during well-known historical events. The failures are explicitly documented with quotes from spokespeople and specific chatbot responses.
• Alternative interpretations: The failures could be interpreted as expected limitations of static LLMs rather than active safety failures, though their integration as default search tools elevates the risk.

During critical political events in July 2024, multiple major AI chatbots failed to provide accurate real-time updates, instead delivering outdated, evasive, or false information. While this did not constitute an active information warfare attack, it highlights a systemic vulnerability in public information ecosystems, as millions of users increasingly rely on autonomous AI systems for breaking news.

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

Threat characteristics

• Imminence: Long-term. Represents an ongoing, systemic challenge with LLM reliability during breaking news rather than an active, imminent national security crisis.
• Autonomy: Full autonomy. The AI chatbots generate and deliver responses to users autonomously without real-time human oversight or validation of the content.
• Novelty: Evolved capability. While LLM hallucinations are well-known, the simultaneous failure of all major commercial models during a sequence of historic national events represents an evolved systemic risk.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or human safety was threatened or compromised by the chatbot failures.
• Information security: Minor. Chatbots provided inaccurate information during critical political events, but this was due to system limitations rather than coordinated foreign information warfare or intelligence compromise.
• Sovereignty: Minor. While the incident involved major political and electoral events, the failures did not disrupt actual government operations, electoral infrastructure, or constitutional processes.
• Economic security: Negligible. No significant threat to financial systems, strategic technology theft, or loss of national technological competitive advantage was reported.
• Societal stability: Minor. The failure to provide accurate real-time information during highly sensitive political events could temporarily confuse the public, but did not lead to civil unrest or systematic rights violations.