#LekkiMassacre: Why Facebook labelled content from October 20 incident ‘false’

Facebook incorrectly labels content relating to an incident between #EndSARS protestors and the Nigerian army as misinformation.

Risk classification

• Primary risk domain: 7 AI system safety, failures, & limitations
• Primary risk subdomain: 7.3 Lack of capability or robustness

The AI system failed to perform reliably under high-stakes real-world conditions, incorrectly flagging authentic crisis documentation as false and exposing the platform to severe operational errors.

Additional risk subdomains

• 3.1 False or misleading information: The system's incorrect flagging of authentic media as false information effectively generated and spread a misleading belief about a real-world event.
• 7.4 Lack of transparency or interpretability: The report highlights that it was not completely clear how Facebook's hybrid system arrived at the conclusion to flag the Lekki-related content.

Causal factors

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

The suppression of legitimate human rights documentation was an unexpected and unintentional outcome of deploying the automated misinformation detection system.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The system is an AI-driven content moderation and labeling tool that interacts with users by displaying warning labels, falling under transparency obligations.

AI system and alleged parties

• AI system: Facebook's signal technology (Meta)
• AI purpose: Content Moderation; Fact Checking
• Behaviour type: Hybrid
• Alleged developer: Facebook
• Alleged deployer: Facebook
• Alleged harmed parties: Facebook users interested in the Lekki Massacre incident, Facebook 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 Minor, indirect Negligible
• Civil rights: direct Minor, indirect Negligible
• Democracy: direct Minor, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Differential treatment

Reported: The report explicitly describes differential treatment of eyewitnesses and protesters whose posts were suppressed compared to other platform users.

Directly caused: The AI system directly flagged and reduced the visibility of posts from #EndSARS protesters and eyewitnesses, restricting their ability to share their experiences.

Inferred additional harm: Thousands of Nigerian citizens sharing authentic crisis documentation likely experienced systematic suppression of their voices relative to state-backed narratives.

Civil rights

Reported: The report explicitly describes the suppression of eyewitness documentation, which directly impacts freedom of expression and the right to report human rights abuses.

Directly caused: The AI system directly restricted the dissemination of crucial human rights information by labeling authentic images of military violence as false.

Inferred additional harm: The suppression of these reports likely hindered international awareness and accountability efforts regarding the human rights violations committed during the protest.

Democracy

Reported: The report explicitly describes how the platform's actions undermined the public's ability to share and access real-time information during a national crisis.

Directly caused: The AI system's incorrect flagging of real-world events as false information delegitimized citizen journalism and public accountability efforts.

Inferred additional harm: This incident likely weakened public trust in social media platforms as viable spaces for democratic mobilization and truth-sharing during crises.

Epistemic

Reported: The report explicitly describes epistemic harm where authentic images of a real-world event were labeled as false, corrupting the shared information environment.

Directly caused: The AI system labeled authentic images of the Lekki toll gate incident as 'False Information', misleading users into believing the military attack did not happen or was fabricated.

Inferred additional harm: This likely caused widespread confusion among millions of social media users in Nigeria and globally, eroding the shared consensus reality of the event.

People affected

• Occurrences reported: 1
• People reportedly harmed: 100
• People reportedly exposed: 10000

Potential causes

Management

• Poor Definition of Misinformation: Facebook management failed to clearly define inaccurate information.
• Inadequate Local Risk Assessment: Insufficient preparation for political crises in African markets.

Technology

• Flawed Automated Matching Algorithm: Algorithmic matching falsely linked real images to debunked posts.
• Inexact AI Content Classification: AI struggled to distinguish context between real and fake reports.

Data Inputs

• Unreliable User Flagging Signals: System relied on user flags which may be coordinated or biased.
• Lack of Contextual Metadata: System lacked real-time local context to verify media authenticity.

Human Factors

• Fact-Checker Bias or Errors: Third-party partners made incorrect assessments of the incident.
• User Inability to Dispute Ratings: Viewers could not challenge incorrect automated false labels.

Process and Methods

• Inadequate Review Escalation: No rapid path to override automated flags during breaking news.
• Flawed Hybrid Moderation Workflow: Integration between AI flagging and human review failed to sync.

Regulatory Environment

• Lack of Government Oversight: No regulatory standards for social media moderation in crises.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear details about the incident, the platforms involved (Facebook and Instagram), the specific real-world event (Lekki toll gate shooting), and how the hybrid AI/human system incorrectly flagged the content. The role of the AI system is explicitly stated, and the resulting harms to information dissemination and human rights reporting are well-documented.
• Ambiguities identified: The exact ratio of AI-driven flags versus human-moderator flags for these specific posts is not fully quantified, though the system is described as a hybrid.
• Alternative interpretations: None. The event is clearly a content moderation failure that suppressed real-world news.

An automated AI content moderation system on Facebook and Instagram incorrectly flagged and suppressed authentic eyewitness media of military violence during Nigeria's #EndSARS protests. This unintentional system failure resulted in substantial societal stability and human rights impacts by restricting real-time crisis reporting and delegitimizing victims' experiences on a massive scale.

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

Threat characteristics

• Imminence: Immediate. The suppression occurred during an active, high-stakes national crisis requiring urgent response to correct the information flow and prevent further public distress.
• Autonomy: Human-supervised. The platform utilizes a hybrid system combining automated AI detection with oversight from human moderators and third-party fact-checkers.
• Novelty: Established threat. Automated content moderation errors and the accidental suppression of legitimate political speech or news during crises are well-documented, recurring issues.

Impact by dimension

• Physical security: Negligible. The incident was a digital content moderation failure and did not present any physical or kinetic threats to critical infrastructure or human safety.
• Information security: Minor. While not a hostile state campaign, the system's failure caused significant information suppression and epistemic harm during an active national crisis.
• Sovereignty: Minor. A foreign private platform's automated decisions impacted domestic political discourse and crisis communication, though it did not directly disrupt core government operations.
• Economic security: Negligible. No significant threats to financial systems, strategic industries, or technological competitive advantage were reported.
• Societal stability: Substantial. The automated system systematically suppressed crucial human rights reporting and eyewitness documentation of military violence, directly impacting freedom of expression and societal trust.