AI Social Scoring
A Threat to Freedom
Artificial intelligence can be used to assign scores to human beings, determining where they can travel, where they can work, and whether they are trustworthy. This is not a future dystopia. It is happening now in China. It needs to be outlawed here.
What Is AI Social Scoring?
Social scoring uses AI to watch, evaluate, and judge human behavior, then assigns consequences based on an algorithmic verdict you cannot see and cannot challenge.
Surveillance
AI monitors behavior across every domain: online activity, financial transactions, physical movement, social relationships, and workplace performance. Every action becomes data.
Evaluation
Machine learning algorithms combine surveillance data into a single score, a numerical representation of your trustworthiness, reliability, and social value.
Consequence
Your score determines access to rights and opportunities: travel, housing, employment, credit, education, and social privileges. Low scores mean exclusion.
It Is Already Here
You may not have a government-issued social credit score, but you have an Uber rating, an Airbnb score, a credit score that now incorporates alternative data, and an algorithmic footprint that determines what you see, what you pay, and what you can access. The mechanics of social scoring are being normalized through consumer apps. The infrastructure is being built voluntarily. The only question is who will control it.
Six Risks of AI Social Scoring
Social scoring is not a single threat. It is a system of interlocking harms that together erode freedom, equality, and human dignity.
Documented Cases of Scoring Harm
These are not hypothetical scenarios. Social scoring systems are already causing measurable harm to individuals and communities.
Travel Bans by Algorithm
China's social credit system has blocked millions of citizens from purchasing high-speed rail tickets or plane flights based on algorithmic scores. The criteria include unpaid fines, jaywalking caught on camera, and social media posts deemed inappropriate.
Algorithmic Sentencing Bias
The COMPAS risk assessment tool used in US courts to predict recidivism was found by ProPublica to falsely flag Black defendants as high risk at nearly twice the rate of white defendants. Judges using these scores gave longer sentences based on algorithmic bias.
Gig Worker Score-Based Exclusion
Platform workers on Uber, DoorDash, and similar services face deactivation based on opaque customer ratings and algorithmic performance scores. Workers report being unable to understand why they were deactivated or how to appeal.
Insurance Scoring Discrimination
Auto insurers using telematics and alternative data sources have been found charging higher premiums to drivers from minority ZIP codes, even when controlling for individual driving records. The algorithms learned geographic bias from historical data.
School Behavior Tracking
ClassDojo and similar classroom behavior apps assign public scores to children for behavior, creating social stratification among students as young as five. Parents report children avoiding peers with low scores to protect their own ratings.
Hiring Algorithm Bias
Amazon developed an AI hiring tool that learned to penalize resumes containing the word 'women's' (as in 'women's chess club captain') and favored resumes with traditionally male-coded language. The system was trained on a decade of hiring data dominated by male hires.
Experts on Social Scoring Risk
Researchers, human rights advocates, and legal scholars who have studied social scoring up close, and what they are warning the public about.
“Social scoring is not a technical innovation. It is a social-control innovation. The same technologies that make it possible also make it possible to build a world where every action is observed, recorded, and evaluated. That world is incompatible with human dignity.”
— Dr. Mireille Hildebrandt, Research Professor on Interdisciplinary Law and Technology, Vrije Universiteit Brussel
“The Chinese social credit system is not an aberration. It is a prototype. The same data infrastructure, the same AI capabilities, and the same corporate actors are building parallel systems in democratic countries, just marketed as convenience, safety, and personalization.”
— Dr. Rogier Creemers, Assistant Professor, Leiden University, leading scholar on China's social credit system
“When an algorithm determines whether you can rent an apartment, get a job, or access credit, and you cannot see the score, cannot challenge the data, and cannot appeal the decision, you are not living in a society with due process. You are living under algorithmic governance.”
— Dr. Virginia Eubanks, Associate Professor of Political Science, University at Albany, author of 'Automating Inequality'
“The most dangerous thing about social scoring is not what it punishes. It is what it teaches people to fear. Once citizens learn that their associations, their words, and even their thoughts may affect their score, they begin to police themselves. The system does not need enforcers; it produces them.”
— Maya Wang, Senior China Researcher, Human Rights Watch
“Every time we accept a user rating, a reputation score, or a personalized algorithmic consequence without demanding transparency and accountability, we are laying another brick in the foundation of social scoring. The question is not whether it will come here. It is already here, wearing different clothes.”
— Shoshana Zuboff, Harvard Business School, author of 'The Age of Surveillance Capitalism'
Six Safeguards Against Social Scoring
Social scoring can be stopped, but only if democratic societies act before the infrastructure becomes too entrenched to dismantle.
Ban Government Social Scoring Systems
National legislation must explicitly prohibit government agencies from operating or contracting social scoring systems that assign numerical ratings to citizens based on behavior, associations, speech, or personal data. The European Parliament has already taken this step. The US must follow.
Mandatory Transparency for All Scoring Algorithms
Any algorithmic system that produces a consequential score, credit, employment, housing, insurance, must disclose the score to the individual, explain the factors that influenced it, and provide a human-reviewable appeals process. Black-box scoring must be illegal.
Prohibit Discriminatory Data Proxies
Algorithms must be prohibited from using proxies for protected characteristics, ZIP code, browsing history, social media activity, neighborhood data, as substitutes for direct discrimination. Independent bias audits must be required before deployment.
Right to Be Unscored
Citizens must have the right to opt out of non-essential scoring systems without penalty. Essential services (credit, housing, employment) may require evaluation, but that evaluation must be conducted transparently and with human oversight, not by opaque algorithm.
Protect Associational Privacy
Scoring systems must be prohibited from incorporating the behavior, scores, or data of family members, friends, colleagues, or associates into an individual's own score. Associational guilt by algorithm must be banned.
Corporate Accountability for Scoring Infrastructure
Technology companies that build, sell, or operate scoring infrastructure must be held liable for harms caused by their systems. This includes facial recognition networks, behavioral tracking platforms, and AI scoring tools sold to governments or corporations.
Why Social Scoring Is Different
Social scoring is not just another privacy issue. It represents a fundamental restructuring of the relationship between the individual and society.
It Reduces People to Numbers
A social score is not a measurement of a person. It is a reduction of a human being to a data point, and that data point determines their access to rights.
It Is Self-Reinforcing
A low score produces worse outcomes, which produces a lower score. The cycle is automated, opaque, and nearly impossible to escape without structural intervention.
It Destroys Solidarity
When your score depends on who you know, human relationships become strategic liabilities. Social scoring turns communities into competitive reputation markets.
Democracy Cannot Coexist with Social Scoring
A society where citizens are watched, scored, and algorithmically rewarded or punished is not a free society. It is a managed society. The infrastructure of social scoring is being built around us, not by government decree, but by the apps we use, the platforms we trust, and the convenience we accept. We must act before the cage is complete.