BlenderBot 3 Cited Dutch Politician as a Terrorist

Meta’s conversational AI BlenderBot 3, when prompted “who is a terrorist,“ responded with an incumbent Dutch politician’s name, who was confused about its association.

Risk classification

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

The AI system generated false and defamatory biographical information about a real individual, leading to potential reputational harm.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The model failed to maintain factual accuracy and robustly distinguish unrelated information, leading to a severe hallucination.

Causal factors

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

The false accusation was generated by Meta's BlenderBot 3 AI system due to an unintentional hallucination or data-mixing error after its public deployment.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The system is a conversational chatbot ('BlenderBot 3') generating text, which falls under the transparency obligations of Risk Level 3.

AI system and alleged parties

• AI system: BlenderBot 3 3 (Meta)
• AI purpose: Chatbot; Question Answering
• Behaviour type: Assistant
• Alleged developer: Meta
• Alleged deployer: Meta
• Alleged harmed parties: Marietje Schaake

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 Minor, 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

Malicious content

Reported: Yes, the chatbot generated a highly toxic and false statement accusing a real person of being a terrorist.

Directly caused: Meta's BlenderBot 3 generated a false statement accusing Marietje Schaake of being a terrorist, which was viewed by her colleague.

Epistemic

Reported: Yes, the report explicitly describes epistemic harm in the form of AI-driven misinformation and fabrication of biographical details.

Directly caused: The AI system fabricated a false biographical claim, stating that according to some governments and international organizations, Marietje Schaake is a terrorist.

Indirectly caused: The spread of this false information on social media highlighted the unreliability of AI chatbots to the public.

People affected

• Occurrences reported: 1
• People reportedly harmed: 1
• People reportedly exposed: 2

Potential causes

Management

• Misaligned Product Positioning: Promoting systems as information sources when not designed for it.

Technology

• Statistical Pattern Prediction: Model joins words based on statistical correlation rather than factual truth.
• Information Combination Error: Model combined two unrelated pieces of info into an incorrect sentence.
• Ability to Freely Generalize: Free generalization untethers outputs from real-world facts.

Data Inputs

• Lack of Online Information: Dearth of online info about specific people causes hallucinations.

Human Factors

• User Misconception of System: Users treat conversational agents as reliable primary information sources.

Process and Methods

• Lack of Output Fact-Checking: System lacks built-in real-time fact-checking before delivering outputs.

Regulatory Environment

• Slim Legal Precedent: Very few existing laws govern AI accuracy and accountability.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide clear, first-hand accounts of the specific prompt and response that occurred, along with Meta's explanation of the technical cause (combining two unrelated pieces of information).

Meta's BlenderBot 3 chatbot falsely labeled Dutch politician Marietje Schaake as a terrorist due to a model hallucination. While the incident caused individual reputational concern, it represents a minor national security threat, illustrating the broader, ongoing challenge of AI-generated misinformation and factual unreliability in public-facing conversational systems.

• Overall national security impact: Minor
• Scope: Multiple nations
• Primary target: No clear primary
• Other affected: Netherlands, United States
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The incident represents an ongoing strategic concern regarding AI factual reliability rather than an active crisis requiring immediate response.
• Autonomy: Full autonomy. The AI system generated and presented the false output directly to the user without prior human review, moderation, or filtering.
• Novelty: Established threat. AI model hallucinations and the generation of false biographical details are well-documented and established issues within large language models.

Impact by dimension

• Physical security: Negligible. The incident involved a conversational chatbot generating text and posed no threat to physical systems, infrastructure, or human safety.
• Information security: Minor. The AI generated false and defamatory information labeling a politician as a terrorist. While not a coordinated information warfare campaign, it highlights risks of AI-driven misinformation.
• Sovereignty: Negligible. No core government operations, electoral systems, or decision-making processes were compromised or targeted by this incident.
• Economic security: Negligible. There was no threat to financial systems, strategic technology theft, or disruption of critical supply chains.
• Societal stability: Minor. The incident caused individual reputational and psychological harm by falsely accusing a person of terrorism, but did not cause large-scale societal instability.