Meta's Discriminatory Ad Machine: How Facebook Enables Housing and Job Bias

In 2022, the Department of Justice reached a landmark settlement with Meta over Facebook's role in facilitating housing discrimination through its advertising platform. The company pledged sweeping reforms. Three years later, independent audits and academic research tell a consistent story: the discrimination continues, driven not by the targeting tools advertisers use but by the algorithmic delivery system Meta controls. The ad platform that reaches nearly 3 billion people is still systematically deciding who sees opportunities based on race, gender, and age — and Meta is still profiting from every discriminatory impression.

The Algorithm Does the Discriminating

The core problem was never just that advertisers could target — or exclude — users by race for housing and employment ads. Even after Meta removed those explicit options, researchers at institutions including Northeastern University and the University of Southern California demonstrated that Facebook's ad delivery algorithm produces discriminatory outcomes on its own. When identical housing ads are submitted without any demographic targeting, the algorithm shows them to racially skewed audiences based on neighborhood data, user behavior patterns, and engagement predictions that serve as proxies for race. A 2025 study found that an apartment listing in a predominantly white neighborhood was delivered to an audience that was 76% white, while an identical listing in a majority-Black neighborhood was delivered to an audience that was 74% Black — without the advertiser specifying any racial preferences.

Job Ads and the Gender Pipeline

Stop guessing about site quality

Get a data-backed score and the exact prompts to fix issues.

The same pattern holds for employment advertising. Researchers submitted identical job advertisements across different industries and found that Facebook's optimization algorithm delivered engineering and technology roles to audiences that were approximately 85% male, while nursing and social work positions were delivered to audiences that were 78% female. This isn't advertisers choosing to discriminate; it's Meta's algorithm reproducing and amplifying societal biases at scale. The effect is a massive digital reinforcement of occupational segregation, delivered through a platform that over 200 million Americans use regularly.

Accountability Gap

Meta earned approximately $115 billion in advertising revenue in 2025. Its investment in ad fairness compliance, by comparison, was less than $30 million — roughly 0.03% of ad revenue. The DOJ settlement included provisions for independent auditing, but the auditors have limited access to Meta's proprietary algorithms and can only test outcomes, not examine the code that produces them. Civil rights organizations including the National Fair Housing Alliance continue to file complaints, and several state attorneys general have opened investigations. For individuals who believe they've been excluded from housing or job opportunities due to Facebook's ad delivery, complaints can be filed with the Department of Housing and Urban Development (HUD) for housing and the Equal Employment Opportunity Commission (EEOC) for employment.

The fundamental tension is this: Meta's ad system is designed to maximize relevance, and in a society marked by structural inequality, 'relevance' as determined by historical data will always reproduce discrimination. Fixing this requires either fundamentally redesigning how ad delivery works or accepting less optimized ad performance in the name of civil rights. So far, Meta has shown no willingness to do either.