A rogue algorithm isn't broken—it's functioning precisely as programmed, yet creating outcomes that were never intended or acknowledged by its creators.
Hidden Impact
It's the scoring system that denies your loan application. The moderation bot that erases your culture. The targeting engine that radicalizes your cousin.
Optimization Gone Wrong
When algorithms prioritize engagement over ethics, they create systems that maximize clicks while potentially minimizing human dignity.
What We Track
Algorithmic Bias & Discrimination
AI systems that disproportionately harm vulnerable communities through flawed design or data.
Opaque Scoring Systems
From credit scores to predictive policing—how invisible mathematical models shape your opportunities and life outcomes.
Social Engineering by Design
When engagement algorithms subtly nudge people into polarization, addiction, or despair.
Optimization Failures
The unintended consequences when algorithms maximize one metric at the expense of human wellbeing.
Featured Algorithmic Failures
Hiring Discrimination
A recruiting algorithm that systematically ghosted every female applicant, perpetuating gender bias in tech industry hiring.
Predictive Policing Errors
A crime prediction model that fabricated entire high-risk neighborhoods based on historical biases in police data.
Cultural Suppression
A content filter that automatically banned indigenous languages, mistakenly flagging them as violations of community standards.
Reinforced Bias
A drone targeting system that re-learned discriminatory patterns despite multiple attempts to remove bias from its training data.
Who We're For
Regulators
Government officials and policy makers seeking accountability and oversight frameworks for algorithmic systems.
Access case studies on regulatory gaps
Find model legislation templates
Connect with technical experts
Researchers
Academic and industry professionals studying AI safety, ethics, and societal impacts.
Access our dataset of documented failures
Contribute to open research initiatives
Find collaboration opportunities
Educators
Teachers and professors creating awareness about algorithmic systems and their societal impacts.
Download classroom materials
Access simplified case studies
Request guest speakers
Engineers
Developers and designers working to prevent unintended consequences in their own systems.
Learn from documented failure patterns
Access ethical design frameworks
Join our technical community
Join the Watchdog Effort
Report Algorithmic Harms
Submit examples of algorithms causing unexpected negative impacts
Study Case Examples
Explore our library of system breakdowns and analysis
Contribute to the Ledger
Help build our public record of documented algorithmic failures
Subscribe to Our Dispatch
Get regular updates from the Algorithmic Underground
By joining our community, you'll gain access to tools, resources, and a network of experts committed to ensuring algorithms serve everyone—not just shareholders. Your contributions help build accountability in our increasingly algorithmic world.
Our Mission
Expose Hidden Systems
Pull back the curtain on algorithmic systems that impact lives without transparency or accountability.
Amplify Affected Voices
Ensure those harmed by algorithmic systems have their experiences documented and heard.
Demand Transparency
Advocate for systems that can be audited, understood, and meaningfully governed by the communities they affect.
Build Inclusive Technology
Support the development of algorithmic systems that serve everyone equitably—not just shareholders or privileged groups.
Latest Case Studies
Healthcare Algorithm Underserved Black Patients
A widely-used healthcare management algorithm systematically assigned lower risk scores to Black patients compared to White patients with the same medical conditions. The algorithm used healthcare costs as a proxy for medical need, but failed to account for systemic barriers that resulted in lower healthcare spending for Black patients despite similar health needs.
Impact: An estimated 4.2 million Black patients received demonstrably less care due to this algorithmic bias.
Facial Recognition Failures Led to False Arrests
Multiple cases have emerged of individuals being wrongfully arrested based on facial recognition algorithms incorrectly matching them to suspects. These systems demonstrate significantly higher error rates for women and people with darker skin tones.
The technology has been deployed by law enforcement with minimal oversight or procedural safeguards against false identifications.
Child Welfare Algorithm Targeted Poor Families
An algorithm designed to identify children at risk of abuse incorrectly flagged families for investigation based primarily on their use of public services—effectively criminalizing poverty rather than identifying actual risk factors for child harm.
Families in low-income neighborhoods were subjected to investigations at rates up to 10 times higher than those in affluent areas with similar actual incident rates.