MIT AI Risk Repository · domain 1: Discrimination & Toxicity

1.1 Unfair discrimination and misrepresentation

Unequal treatment of individuals or groups by AI, often based on race, gender, or other sensitive characteristics, resulting in unfair outcomes and representation of those groups.

Risk entries
83
Frameworks citing it
12
Recorded incidents
118
Incidents since 2020
58
Causal entity (risk entries)
Causal entity (risk entries) 58 0 AI: 58 AI 58 Other: 13 Other 13 Human: 11 Human 11 Not coded: 1 Not coded 1
Causal entity (risk entries)
LabelValue
AI58
Other13
Human11
Not coded1
Intent (risk entries)
Intent (risk entries) 64 0 Unintentional: 64 Unintentional 64 Other: 16 Other 16 Intentional: 2 Intentional 2 Not coded: 1 Not coded 1
Intent (risk entries)
LabelValue
Unintentional64
Other16
Intentional2
Not coded1
Timing (risk entries)
Timing (risk entries) 50 0 Post-deployment: 50 Post-deployment 50 Other: 19 Other 19 Pre-deployment: 13 Pre-deployment 13 Not coded: 1 Not coded 1
Timing (risk entries)
LabelValue
Post-deployment50
Other19
Pre-deployment13
Not coded1
Recorded incidents per yearIncident date; current year partial
Recorded incidents per year 19 0 2012: 5 2012 5 2013: 2 2013 2 2014: 2 2014 2 2015: 6 2015 6 2016: 13 2016 13 2017: 9 2017 9 2018: 9 2018 9 2019: 8 2019 8 2020: 19 2020 19 2021: 8 2021 8 2022: 10 2022 10 2023: 11 2023 11 2024: 7 2024 7 2025: 3 2025 3
Recorded incidents per year
LabelValue
20125
20132
20142
20156
201613
20179
20189
20198
202019
20218
202210
202311
20247
20253
Entries by levelRisk categories, subcategories and additional evidence coded to this subdomain
Entries by level 61 0 Risk Category: 22 Risk Category 22 Risk Sub-Category: 61 Risk Sub-Category 61
Entries by level
LabelValue
Risk Category22
Risk Sub-Category61
  • Denying people the opportunity to self-identify

    complex and non-traditional ways in which humans are represented and classified automatically, and often at the cost of autonomy loss... such as categorizing someone who identifies as non-binary into...

    Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction (Shelby2023) · AI · Unintentional · Post-deployment

  • Reifying essentialist categories

    algorithmic systems that reify essentialist social categories can be understood as when systems that classify a person’s membership in a social group based on narrow, socially constructed criteria tha...

    Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction (Shelby2023) · AI · Unintentional · Post-deployment

  • Allocative Harms

    "These harms occur when a system withholds information, opportunities, or resources [22] from historically marginalized groups in domains that affect material well-being [146], such as housing [47], e...

    Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction (Shelby2023) · AI · Unintentional · Post-deployment

  • Opportunity loss

    Opportunity loss occurs when algorithmic systems enable disparate access to information and resources needed to equitably participate in society, including the withholding of housing through targeting...

    Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction (Shelby2023) · AI · Unintentional · Post-deployment

  • Economic loss

    Financial harms [52, 160] co-produced through algorithmic systems, especially as they relate to lived experiences of poverty and economic inequality... demonetization algorithms that parse content tit...

    Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction (Shelby2023) · AI · Unintentional · Post-deployment

  • Fairness & Bias

    "The potential for AI systems to make decisions that systematically disadvantage certain groups or individuals. Bias can stem from training data, algorithmic design, or deployment practices, leading t...

    AI Risk Profiles: A Standards Proposal for Pre-Deployment AI Risk Disclosures (Sherman2023) · AI · Unintentional · Other

  • Bias, Stereotypes, and Representational Harms

    "Generative AI systems can embed and amplify harmful biases that are most detrimental to marginalized peoples."

    Evaluating the Social Impact of Generative AI Systems in Systems and Society (Solaiman2023) · AI · Unintentional · Other

  • Inequality, Marginalization, and Violence

    "Generative AI systems are capable of exacerbating inequality, as seen in sections on 4.1.1 Bias, Stereotypes, and Representational Harms and 4.1.2 Cultural Values and Sensitive Content, and Disparate...

    Evaluating the Social Impact of Generative AI Systems in Systems and Society (Solaiman2023) · Other · Other · Post-deployment

  • Discriminatory and exclusionary language

    -

    Emerging Risks and Mitigations for Public Chatbots: LILAC v1 (Stanley2024) · AI · Other · Post-deployment

  • Biased statements and recommendations

    "The chatbot gives information that, while not obviously false or harmful, could lead to biased decision-making."

    Emerging Risks and Mitigations for Public Chatbots: LILAC v1 (Stanley2024) · AI · Unintentional · Other

  • Fairness

    "The general principle of equal treatment requires that an AI system upholds the principle of fairness, both ethically and legally. This means that the same facts are treated equally for each person u...

    Sources of Risk of AI Systems (Steimers2022) · AI · Unintentional · Post-deployment

  • Unfairness and discrinimation

    "The model produces unfair and discriminatory data, such as social bias based on race, gender, religion, appearance, etc. These contents may discomfort certain groups and undermine social stability an...

    Safety Assessment of Chinese Large Language Models (Sun2023) · AI · Other · Post-deployment

  • Discrimination

    This is the risk of an ML system encoding stereotypes of or performing disproportionately poorly for some demographics/social groups.

    The Risks of Machine Learning Systems (Tan2022) · AI · Unintentional · Post-deployment

  • Risks from models and algorithms (Risks of bias and discrimination)

    "During the algorithm design and training process, personal biases may be introduced, either intentionally or unintentionally. Additionally, poor-quality datasets can lead to biased or discriminatory...

    AI Safety Governance Framework (TC2602024) · Human · Other · Pre-deployment

  • Bias

    "A systematic error, a tendency to learn consistently wrongly."

    An Exploratory Diagnosis of Artificial Intelligence Risks for a Responsible Governance (Teixeira2022) · AI · Unintentional · Pre-deployment

  • Discrimination

    "The creation, perpetuation or exacerbation of inequalities and biases at a large-scale."

    A Taxonomy of Systemic Risks from General-Purpose AI (Uuk2025) · Other · Other · Post-deployment

  • Incomplete or biased training data

    "Incomplete or biased training data can lead to discriminatory AI outputs."

    A Taxonomy of Systemic Risks from General-Purpose AI (Uuk2025) · Other · Unintentional · Post-deployment

  • Social stereotypes and unfair discrmination

    "Perpetuating harmful stereotypes and discrimination is a well-documented harm in machine learning models that represent natural language (Caliskan et al., 2017). LMs that encode discriminatory langua...

    Ethical and social risks of harm from language models (Weidinger2021) · AI · Unintentional · Other

  • Exclusionary norms

    "In language, humans express social categories and norms. Language models (LMs) that faithfully encode patterns present in natural language necessarily encode such norms and categories...such norms an...

    Ethical and social risks of harm from language models (Weidinger2021) · AI · Unintentional · Other

  • Promoting harmful stereotypes by implying gender or ethnic identity

    "A conversational agent may invoke associations that perpetuate harmful stereotypes, either by using particular identity markers in language (e.g. referring to “self” as “female”), or by more general...

    Ethical and social risks of harm from language models (Weidinger2021) · AI · Unintentional · Post-deployment

  • Social stereotypes and unfair discrimination

    "The reproduction of harmful stereotypes is well-documented in models that represent natural language [32]. Large-scale LMs are trained on text sources, such as digitised books and text on the interne...

    Taxonomy of Risks posed by Language Models (Weidinger2022) · AI · Unintentional · Other

  • Exclusionary norms

    "In language, humans express social categories and norms, which exclude groups who live outside of them [58]. LMs that faithfully encode patterns present in language necessarily encode such norms."

    Taxonomy of Risks posed by Language Models (Weidinger2022) · AI · Unintentional · Other

  • Promoting harmful stereotypes by implying gender or ethnic identity

    "CAs can perpetuate harmful stereotypes by using particular identity markers in language (e.g. referring to “self” as “female”), or by more general design features (e.g. by giving the product a gender...

    Taxonomy of Risks posed by Language Models (Weidinger2022) · AI · Unintentional · Post-deployment

  • Unfair representation

    "Mis-, under-, or over-representing certain identities, groups, or perspectives or failing to represent them at all (e.g. via homogenisation, stereotypes)"

    Sociotechnical Safety Evaluation of Generative AI Systems (Weidinger2023) · AI · Unintentional · Post-deployment

  • Erosion of trust in public information

    "Eroding trust in public information and knowledge"

    Sociotechnical Safety Evaluation of Generative AI Systems (Weidinger2023) · Human · Other · Post-deployment