AIPolicyTracker
Menu

MIT AI Risk Repository · Risk Category · 18.01.00

Representation & Toxicity Harms

Open the MIT Risk Navigator All entries from Weidinger2023

Description

"AI systems under-, over-, or misrepresenting certain groups or generating toxic, offensive, abusive, or hateful content"

From Sociotechnical Safety Evaluation of Generative AI Systems (Weidinger2023), as extracted by the MIT AI Risk Repository (CC BY 4.0).

Classification

Subdomain
1.0
Causal entity
AI

How other frameworks describe this risk

  • Stereotyping

    A Collaborative, Human-Centred Taxonomy of AI, Algorithmic, and Automation Harms (Abercrombie2024)

  • Impacts of AI (Bias)

    Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

  • -

    A Closer Look at the Existing Risks of Generative AI: Mapping the Who, What, and How of Real-World Incidents (Li2025)

  • Discrimination, Exclusion and Toxicity

    Ethical and social risks of harm from language models (Weidinger2021)

  • Ethical AI Risks

    Governance of artificial intelligence: A risk and guideline-based integrative framework (Wirtz2022)

  • Discrimination/Bias (Discriminatory Activities)

    AI Risk Categorization Decoded (AIR 2024): From Government Regulations to Corporate Policies (Zeng2024)

  • Unfairness and Bias

    SafetyBench: Evaluating the Safety of Large Language Models with Multiple Choice Questions (Zhang2023)

Other entries from Weidinger2023