Chinese Insurer Ping An Employed Facial Recognition to Determine Customers’ Untrustworthiness, Which Critics Alleged to Likely Make Errors and Discriminate

Customers’ untrustworthiness and unprofitability were reportedly determined by Ping An, a large insurance company in China, via facial-recognition measurements of micro-expressions and body-mass indices (BMI), which critics argue was likely to make mistakes, discriminate against certain ethnic groups, and undermine its own industry.

Risk classification

• Primary risk domain: 1 Discrimination & Toxicity
• Primary risk subdomain: 1.1 Unfair discrimination and misrepresentation

The AI system is used to discriminate against individuals based on BMI and is projected to unfairly exclude low-income individuals and ethnic minorities like Uyghurs from insurance coverage.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The report strongly questions the scientific validity and technical capability of using facial micro-expressions to accurately detect deception.

Causal factors

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

The risk arises from the active deployment of Ping An's facial recognition AI to make real-world decisions about customer insurance eligibility and premiums.

EU AI Act risk tier

• Risk tier: 2 High Risk

Risk Level 2. High Risk: The report describes 'facial recognition software' used to evaluate individuals for insurance eligibility, which falls under biometric identification and essential private services.

AI system and alleged parties

• AI system: Ping An facial recognition software
• AI purpose: Trustworthiness Scoring; Underwriting
• Behaviour type: Assistant
• Alleged developer: Ping An
• Alleged deployer: Ping An
• Alleged harmed parties: Ping An customers, Chinese minority groups

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

Differential treatment

Reported: Yes, the report explicitly describes differential treatment where individuals with higher BMIs are charged extra, and poor or ethnic minority groups are targeted for exclusion.

Directly caused: The AI system directly charges extra premiums to individuals based on their body-mass index and well-being metrics.

Indirectly caused: The system's focus on avoiding claimants indirectly leads to the exclusion of low-income individuals and ethnic minorities like Uyghurs who are statistically more likely to make claims.

Inferred additional harm: Widespread differential treatment likely affected millions of Ping An's customer base in China, systematically raising costs or denying services to marginalized groups.

Civil rights

Reported: Yes, the report describes potential discrimination against ethnic minorities like Uyghurs, which constitutes a violation of civil and human rights.

Indirectly caused: The deployment of the system indirectly facilitates systemic discrimination against the Uyghur Muslim minority by using poverty and claim-likelihood as proxies for exclusion.

Inferred additional harm: The systematic exclusion of marginalized groups from essential services like health insurance likely violates basic civil rights on a large scale across China.

Privacy

Reported: Yes, the report mentions the use of facial recognition software to search for micro-expressions and gauge body-mass index.

Directly caused: The system analyzes customers' facial features and micro-expressions to make determinations about their truthfulness and health.

Inferred additional harm: The collection and algorithmic analysis of sensitive biometric data (facial micro-expressions and BMI) without robust regulatory protections likely constitutes a widespread invasion of privacy for millions of applicants.

People affected

• Occurrences reported: 1

Potential causes

Management

• Prioritizing Profit Over Fairness: Management focuses on avoiding expensive claims at the expense of equity.
• Erosion of Risk-Pooling Principles: Redefining insurance to exclude those who actually need the coverage.

Technology

• Flawed Micro-expression Detection: Using unproven science to determine truthfulness from facial micro-expressions.
• Inaccurate BMI and Health Estimation: Using facial recognition to estimate BMI and penalize health premiums.

Data Inputs

• Unrepresentative Training Data: Algorithm trained on laboratory lies which do not match real-world behavior.
• Omission of Undetected Lies: Training data lacks real-world lies that were never caught or discovered.
• Proxy Data for Marginalized Groups: Claims data correlates poverty and minority status with high insurance risks.

Human Factors

• Overreliance on Algorithmic Decisions: Insurers trust AI claims of truthfulness without human verification.
• Behavioral Masking by Good Liars: Good liars control expressions, causing the AI to misidentify dishonesty.

Process and Methods

• Unfair Risk-Profiling Methods: Profiling customers based on demographic and physical traits to deny coverage.
• Lack of Scientific Validation: Deploying truth-detection methods without robust scientific proof of efficacy.

Regulatory Environment

• Absence of Anti-Discrimination Laws: Lack of regulatory mandates preventing insurers from excluding high-risk groups.
• Repeal of Universal Coverage Mandates: Weakening of laws that require insurers to cover all individuals equally.

Information quality

• Classification confidence: High
• Reason for confidence: The report clearly outlines the AI system's deployment by Ping An, its intended functions (micro-expression and BMI analysis), and the specific socio-economic risks of discrimination. While it is an analytical piece, the core facts of the deployment are clearly stated.
• Ambiguities identified: The exact scale of deployment and the precise number of customers affected are not quantified in the text.
• Alternative interpretations: The system could be viewed as a standard fraud-detection and risk-assessment tool rather than an inherently discriminatory instrument.

Ping An deployed an AI-driven biometric system using facial recognition and micro-expressions for insurance underwriting. While a commercial application, it poses substantial risks to societal stability and human rights by enabling systemic discrimination against marginalized populations, specifically low-income groups and ethnic minorities like the Uyghurs.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: China
• Alleged perpetrator: Ping An

Threat characteristics

• Imminence: Long-term. Represents an ongoing corporate capability and strategic human rights concern rather than an immediate national security crisis.
• Autonomy: Human-supervised. The AI system assists human underwriters in making decisions rather than operating with full, unreviewed autonomy.
• Novelty: Evolved capability. Represents an evolution of facial recognition and biometric profiling technologies applied to commercial underwriting.

Impact by dimension

• Physical security: Negligible. No physical security threats, kinetic attacks, or critical infrastructure compromise reported in this commercial insurance deployment.
• Information security: Negligible. No evidence of information warfare, intelligence compromise, or systematic state-level disinformation operations.
• Sovereignty: Negligible. The commercial deployment does not threaten state authority, territorial control, or core government operations.
• Economic security: Negligible. No threat to strategic industries, financial system stability, or national technological competitive advantage.
• Societal stability: Substantial. Facilitates systemic discrimination and potential human rights violations against marginalized groups, including ethnic minorities like Uyghurs, undermining societal stability.