In a landmark decision, the Netherlands Institute for Human Rights (NIHR) has ruled that the advertising algorithm used by Facebook, operated by Meta Platforms Ireland Ltd in Europe, engages in indirect gender discrimination when displaying job advertisements.
According to the NIHR’s decision of 18 February 2025 (Decision 2025-17), research by campaign organisation Global Witness found that in 2022-23 certain job ads shown on Facebook were overwhelmingly displayed to one gender. For example:
- A vacancy for “receptionist” was shown to female users in 96% of cases in 2022 and 97% in 2023.
- A vacancy for “mechanic” was shown to male users in 96% of cases in both years.
The NIHR held that the algorithm’s effect, even though it may appear gender-neutral, was to perpetuate gender stereotypes, thus constituting indirect discrimination under Dutch and EU equal-treatment law.
Why this matters for business and governance
The ruling means that algorithmic practices will be held to the same non-discrimination standards as offline business practices. This is especially important for companies that run digital advertising platforms or use algorithmic targeting.
It highlights that algorithms are not a regulatory “safe harbor”—even”if users actively click and engage, the design of the system may still produce discriminatory outcomes. It emphasises that duty of care exists: service providers cannot simply deploy machine-learning systems and say, “We didn’t intend discrimination”; they must monitor and mitigate bias.
For governance professionals, compliance frameworks must extend beyond input checks and documented design: impact assessment, auditing, and transparency of outcomes are increasingly critical.
Implications for stakeholders
Advertisers and platforms—platforms that segment, target, or optimise ads based on user characteristics (gender, age, and interests) must assess whether their targeting or auction mechanism systematically favours or excludes protected groups. Even if you disable gender targeting, the algorithmic downstream effect could potentially recreate skewed distributions.
Human resources and recruiting firms – When using social media ads or programmatic job postings, HR leaders must recognise that skewed reach may result in narrower applicant pools or reinforce traditional job-gender stereotypes (e.g., “male mechanic” or “female receptionist”)—potentially increasing the risk of challenge under equal-opportunity law.
Regulators and compliance teams – This sets a precedent: human rights or equality bodies are willing to treat algorithms as “services” under non-discrimination law. After the ruling, organisations must build frameworks to audit algorithmic fairness, document mitigation efforts, and demonstrate that any disparate impacts are objectively justified.
Brand reputation and ESG – For firms committed to gender equity, the ruling underscores that fairness must cover not just hiring practices but also the tools used to hire, advertise, and engage talent or audience. Legacy brands will increasingly need to prove that their digital operations align with diversity, equity & inclusion (DEI) goals.
What’s next?
The NIHR’s decision is non-binding, meaning Meta is not legally forced to comply, but the principle is clear. In fact, similar regulatory scrutiny is underway in other jurisdictions.
Key questions now for businesses include:
- How to measure and report the disparate impact of ad-delivery systems?
- What alternative algorithms or safeguards (e.g., randomisation, fairness constraints) are technically feasible and proportionate?
- How to document and justify any targeting approach that produces different reach by gender or other protected characteristic?
- How to integrate algorithmic fairness into existing governance frameworks—including board-level oversight, risk registers, and external audits?
The Netherlands’ ruling marks a watershed moment: the era when algorithmic decision-making was too opaque to touch is clearly ending. For business leaders, especially in recruitment, advertising and platform operations, the message is unambiguous: digital systems must reflect the same equality obligations as traditional services.
The effect of your algorithms matters more than avoiding the “intention” to discriminate. And if your technology favours or excludes one gender from seeing job opportunities or advertisements, the law may soon treat that as unacceptable.
