Facebook Political Ad Delivery Algorithms Inferred Users' Political Alignment, Inhibiting Political Campaigns' Reach

Facebook's political ad delivery system reportedly differentiated the price of user reach based on their inferred political alignment, inhibiting political campaigns' ability to reach voters with diverse political views, which allegedly reinforces political polarization and creates informational filter bubbles.

Risk classification

• Primary risk domain: 3 Misinformation
• Primary risk subdomain: 3.2 Pollution of information ecosystem and loss of consensus reality

The algorithm preferentially exposes users to political ads matching their inferred alignment, creating informational filter bubbles and contributing to political polarization.

Additional risk subdomains

• 1.1 Unfair discrimination and misrepresentation: The report notes that the ad delivery algorithms can selectively deliver ads in ways that lead to demographic skews along race and gender lines.

Causal factors

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

The polarization and pricing differentiation are unintentional side effects of the deployed ad delivery algorithm optimizing for relevance.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The report describes an AI system with significant implications for fundamental rights (non-discrimination along race and gender lines) and public interests (political polarization and democratic elections).

AI system and alleged parties

• AI system: Facebook's ad delivery algorithms (Meta)
• AI purpose: Ad Delivery; Content Recommendation
• Behaviour type: Autonomous
• Alleged developer: Facebook
• Alleged deployer: Facebook
• Alleged harmed parties: political campaigns, Facebook users

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Substantial. 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 Minor, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Minor, indirect Substantial
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Minor, indirect Minor
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Substantial, indirect Severe
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: The report notes that Facebook's algorithms differentiate the price of reaching users, which can increase costs for political campaigns, but does not quantify specific financial losses.

Directly caused: The algorithm directly increased the cost of ad delivery for campaigns targeting users with non-aligned political views, though specific monetary values are not reported.

Inferred additional harm: It is likely that political campaigns, especially those with smaller budgets, incurred higher advertising costs or reduced reach per dollar spent due to the pricing differentiation, though the total financial impact is unquantified.

Differential treatment

Reported: The report explicitly describes differential treatment where the algorithm differentiates the price and delivery of ads based on users' inferred political alignments and demographic characteristics.

Directly caused: Facebook's ad delivery algorithm preferentially delivered ads to users based on their inferred political alignment, charging different prices to reach different users.

Indirectly caused: The algorithm selectively delivered ads within target audiences in ways that led to demographic skews along race and gender lines without the advertisers' knowledge.

Inferred additional harm: It is likely that millions of Facebook users experienced highly differentiated information environments and ad pricing structures based on their demographic and political profiles.

Democracy

Reported: The report explicitly describes how the algorithm's selective delivery inhibits political campaigns' ability to reach voters with diverse views, contributing to political polarization.

Directly caused: The algorithm preferentially exposed users to political advertising matching their inferred alignment, creating informational filter bubbles.

Indirectly caused: This selective exposure and pricing differentiation restricts the free flow of diverse political viewpoints, potentially undermining democratic debate and polarizing the electorate.

Inferred additional harm: The systemic polarization of voters and the creation of filter bubbles likely weakens democratic cohesion and the functioning of democratic processes at a national scale.

Epistemic

Reported: The report explicitly describes epistemic harm in the form of informational filter bubbles that restrict users' exposure to diverse political viewpoints.

Directly caused: The algorithm preferentially delivered content that aligned with users' existing views, reinforcing their beliefs and shielding them from opposing perspectives.

Indirectly caused: This contributes to a fragmented information ecosystem where users lack a shared consensus reality, accelerating political polarization.

Inferred additional harm: Widespread exposure to highly personalized political ad delivery likely deepens societal divisions and reduces the public's shared understanding of political issues.

People affected

• Occurrences reported: 1

Potential causes

Management

• Prioritizing Relevance Over Diversity: Platform prioritizes user engagement over ensuring diverse political views.

Technology

• Algorithmic Relevance Optimization: Algorithm prioritizes users estimated to be most relevant, causing skews.
• Dynamic Pricing Based on Alignment: Differentiates ad cost based on inferred political alignment of users.

Data Inputs

• Inferred Political Affiliations: Using inferred user profiles to target and deliver political ads.

Human Factors

• Advertiser Ignorance of Skew: Campaigns cede control to algorithms without knowing delivery is skewed.

Process and Methods

• Opaque Ad Performance Reporting: Platform reports hide algorithmic delivery skews from political campaigns.
• Ineffective Targeting Restrictions: Restricting targeting tools fails to stop algorithmic delivery bias.

Regulatory Environment

• Lack of Public Transparency Rules: No regulations forcing platforms to disclose algorithmic delivery metrics.

Information quality

• Classification confidence: High
• Reason for confidence: The report is a peer-reviewed academic study detailing empirical experiments on Facebook's ad platform. The methodology, findings, and implications are clearly laid out, providing high-quality, reliable information about the algorithm's behavior and its societal impacts.
• Ambiguities identified: The exact proprietary mechanics of Facebook's ad delivery and pricing algorithms are not fully public, requiring researchers to infer their behavior through black-box testing.
• Alternative interpretations: The pricing differentiation could be interpreted purely as market-driven optimization for user relevance rather than an intentional effort to polarize or restrict reach.

Facebook's ad delivery algorithm unintentionally polarizes the electorate and restricts political campaigns' reach by differentiating ad costs based on inferred political alignment. This systemic behavior creates informational filter bubbles and demographic skews, presenting a substantial long-term challenge to democratic processes and societal stability.

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

Threat characteristics

• Imminence: Long-term. Represents an ongoing strategic concern regarding algorithmic bias and polarization rather than an active, immediate crisis.
• Autonomy: Full autonomy. The ad delivery algorithm makes real-time distribution and pricing decisions autonomously without human intervention or oversight.
• Novelty: Evolved capability. Represents an advanced, systemic evolution of known algorithmic targeting issues, specifically highlighting invisible pricing biases.

Impact by dimension

• Physical security: Negligible. The incident involves digital ad delivery algorithms with no physical real-world impact, kinetic targeting, or threat to critical infrastructure.
• Information security: Substantial. The algorithm systematically manipulates the information ecosystem by creating informational filter bubbles and political polarization, though this is an unintentional platform mechanic rather than a hostile foreign campaign.
• Sovereignty: Substantial. The ad delivery mechanism interferes with the equitable execution of democratic processes by restricting campaigns' ability to reach diverse voters during a national election.
• Economic security: Minor. Causes minor economic impacts by artificially inflating ad delivery costs for political campaigns, particularly affecting campaigns with smaller budgets.
• Societal stability: Substantial. Fosters systemic political polarization and introduces demographic skews along race and gender lines, undermining social cohesion and equitable information access.