Amazon Allegedly Tweaked Search Algorithm to Boost Its Own Products

Amazon tweaked product-search algorithm to boost and guide customers towards more profitable in-house products instead of showing mainly most-relevant and best-selling listings, which its internal engineers and lawyers alleged to violate company’s best-for-customer principle.

Risk classification

• Primary risk domain: 6 Socioeconomic & Environmental
• Primary risk subdomain: 6.1 Power centralization and unfair distribution of benefits

Amazon used its dominant platform and proprietary search algorithm to favor its own products and squeeze out independent merchants, centralizing market power and resources.

Additional risk subdomains

• 4.3 Fraud, scams, and targeted manipulation: The algorithm manipulated search results and buy box selections to steer customers toward more expensive products for Amazon's financial gain.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The risk was caused by Amazon executives intentionally directing engineers to modify the deployed search algorithm to prioritize profitability over relevance.

EU AI Act risk tier

• Risk tier: 4 Minimal or No Risk

Minimal or No Risk: The product search and ranking algorithm is a commercial search tool that does not fall under the prohibited or high-risk categories defined in the EU AI Act.

AI system and alleged parties

• AI system: A9 search algorithm (Amazon)
• AI purpose: Content Search; Product Recommendation
• Behaviour type: Tool
• Alleged developer: Amazon
• Alleged deployer: Amazon
• Alleged harmed parties: small businesses on Amazon, Amazon Customers

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 Substantial, indirect Substantial
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Substantial, indirect Substantial
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: The report describes financial losses to consumers who paid more for products and to third-party sellers who lost sales or paid high fees.

Directly caused: Consumers paid an average of 20 percent more, or about $7.88 extra per product, when purchasing from the algorithm-selected buy box.

Indirectly caused: Sellers like BareBones WorkWear closed a warehouse and call center, and Beauty Bridge saw a 30 percent decline in sales before paying 15 percent FBA fees.

Inferred additional harm: Given Amazon's 300 million active accounts, aggregate consumer losses and seller fees likely total hundreds of millions of dollars.

Differential treatment

Reported: The report explicitly describes differential treatment of third-party sellers compared to Amazon's own products and FBA partners.

Directly caused: Amazon's search and buy box algorithms favored Amazon's own products and FBA sellers by omitting shipping costs in their rankings.

Indirectly caused: Non-FBA sellers were relegated to lower search rankings and excluded from the buy box, forcing them to pay high fees to compete.

Inferred additional harm: Millions of independent merchants on Amazon's platform likely faced systemic disadvantages compared to Amazon's private labels.

People affected

• Occurrences reported: 1
• People reportedly harmed: 2
• People reportedly exposed: 301000000

Potential causes

Management

• Prioritization of profit over trust: Executives prioritized retail profitability over customer-centric principles.
• Reorganization of search team: A9 search team was moved under retail management, reducing independence.
• Bypassing of legal constraints: Management pushed for indirect profit proxies to avoid antitrust fines.

Technology

• Algorithmic self-preferencing: Algorithm modified to favor Amazon own-brand and FBA-partner products.
• Profitability proxy variables: Engineers used indirect variables to boost profitability without direct metrics.
• Removal of relevance default sort: Default sort changed from relevance to featured to mask profitability bias.

Data Inputs

• Omission of shipping cost data: Algorithm omitted shipping costs for Amazon/FBA items in price comparisons.
• Integration of profit data: Engineers imported contribution profit data to train and test the algorithm.

Human Factors

• Pressure on search engineers: Retail executives pressured A9 engineers to prioritize profit over relevance.
• Consumer trust in Buy Box neutrality: Shoppers assumed default recommendations represented the cheapest or best deals.

Process and Methods

• Profitability added to dashboard: Inclusion of profit metrics in Weblab incentivized biased algorithm designs.
• Biased Buy Box selection logic: Selection criteria favored FBA sellers over cheaper independent merchants.
• Lack of algorithmic transparency: Secret algorithms prevented external auditing of self-preferencing behavior.

Regulatory Environment

• Absence of algorithmic oversight: Lack of regulation allowed hidden manipulation of search results.
• Unclear antitrust frameworks: Existing laws struggled to address digital platform self-preferencing.

Information quality

• Classification confidence: High
• Reason for confidence: The reports from WSJ, ProPublica, and The Verge provide detailed, investigative accounts of Amazon's search algorithm changes, including internal conflicts, specific testing metrics, and empirical data from testing 250 products.
• Ambiguities identified: Amazon's official statements deny changing criteria to include profitability directly, creating a conflict with the internal sources' accounts of using profitability proxies.
• Alternative interpretations: Amazon argues that the algorithm is optimized for customer trust and long-term value, and that considering profitability is standard business practice rather than anti-competitive manipulation.

Amazon modified its product search algorithm to prioritize its own profitable products, causing financial harm to independent sellers and consumers. While raising significant antitrust and regulatory concerns, the incident remains a commercial dispute with minor national security implications manageable through standard legal and economic frameworks.

• Overall national security impact: Minor
• Scope: Multiple nations
• Primary target: No clear primary
• Other affected: Multiple nations
• Alleged perpetrator: Amazon

Threat characteristics

• Imminence: Long-term. The algorithmic changes represent a long-term commercial strategy and ongoing market practice rather than an active crisis.
• Autonomy: Human-supervised. The search algorithm operates autonomously to rank and serve millions of products, but is designed and adjusted under close human supervision.
• Novelty: Established threat. Algorithmic self-preferencing and platform bias are well-known commercial practices with existing regulatory precedents.

Impact by dimension

• Physical security: Negligible. No physical security threats, critical infrastructure disruption, or kinetic impacts were reported.
• Information security: Negligible. No intelligence compromise, state-sponsored information operations, or classified data theft occurred.
• Sovereignty: Negligible. The incident did not impact state authority, electoral systems, or core government decision-making.
• Economic security: Minor. Represents commercial market manipulation causing financial loss to independent sellers and consumers, manageable through antitrust and regulatory frameworks rather than national security apparatus.
• Societal stability: Negligible. No significant threats to societal stability, civil liberties, or human rights were identified.