MIT AI Risk Repository

Browse AI risks

Name: MIT AI Risk Repository: AI Risk Database (browseable copy)
Creator: MIT AI Risk Initiative
License: https://creativecommons.org/licenses/by/4.0/

237 risk entries extracted from 74 frameworks, coded by domain, subdomain, causal entity, intent and timing. Filter, then export the current selection with its licence and citation attached.

Export CSV Export JSON

237 entries · page 1 of 5

62.34.00 · Risk Category

Impacts of AI (Bias)

—

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

Domain 1 1.0 Not coded Not coded Not coded
62.10.01 · Risk Sub-Category

Direct Harm Domains (legal and rights-related harms)

Discrimination and bias

—

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

Domain 1 1.1 Not coded Not coded Not coded
23.01.01 · Risk Sub-Category

Violent crimes

Mass violence

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.01.02 · Risk Sub-Category

Violent crimes

Murder

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.01.03 · Risk Sub-Category

Violent crimes

Physical assault against a person

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.01.04 · Risk Sub-Category

Violent crimes

Violent domestic abuse

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.01.05 · Risk Sub-Category

Violent crimes

Terror (Terror groups, Terror actors, Terrorist actions)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.02.01 · Risk Sub-Category

Non-violent crimes

Theft

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.02.02 · Risk Sub-Category

Non-violent crimes

Modern slavery and human trafficking

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.02.03 · Risk Sub-Category

Non-violent crimes

Child abuse (non-sexual)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.02.04 · Risk Sub-Category

Non-violent crimes

Property damage

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.02.05 · Risk Sub-Category

Non-violent crimes

Financial crime (Scams, Fraud)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.02.06 · Risk Sub-Category

Non-violent crimes

Sale, purchase or use of illegal, highly regulated or unregistered items (Dangerous substances, Dangerous weapons, Prohibited wildlife, Fake goods)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.03.01 · Risk Sub-Category

Sex-related crimes

Sexual assault (e.g., rape)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.03.02 · Risk Sub-Category

Sex-related crimes

Sexual harassment

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.03.03 · Risk Sub-Category

Sex-related crimes

Sex trafficking

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.03.04 · Risk Sub-Category

Sex-related crimes

Prostitution

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.03.05 · Risk Sub-Category

Sex-related crimes

Intimate-image based abuse

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.03.06 · Risk Sub-Category

Sex-related crimes

Indecent exposure (i.e. showing sexual body parts or "flashing")

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.04.01 · Risk Sub-Category

Child sexual exploitation

Inappropriate/harmful relationships with children. This includes adults forming romantic relationships with children or grooming them

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.04.02 · Risk Sub-Category

Child sexual exploitation

Sexual abuse of children, including the sexualisation of children

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.04.03 · Risk Sub-Category

Child sexual exploitation

Child Sexual Abuse Material (CSAM). This includes erotic materials involving children

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.05.01 · Risk Sub-Category

Indiscriminate weapons (CBRNE)

CBRNE (all)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.05.02 · Risk Sub-Category

Indiscriminate weapons (CBRNE)

Chemical weapons (ex: nerve gas)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.05.03 · Risk Sub-Category

Indiscriminate weapons (CBRNE)

Biological weapons (ex: anthrax)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.05.04 · Risk Sub-Category

Indiscriminate weapons (CBRNE)

Radiological weapons (ex: dirty bombs)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.05.05 · Risk Sub-Category

Indiscriminate weapons (CBRNE)

Nuclear weapons (ex: fission or fusion bombs)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.05.06 · Risk Sub-Category

Indiscriminate weapons (CBRNE)

High-yield explosive weapons (ex: cluster munitions)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.06.01 · Risk Sub-Category

Suicide & self-harm

Suicide

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.06.02 · Risk Sub-Category

Suicide & self-harm

Self-harm

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.06.03 · Risk Sub-Category

Suicide & self-harm

Eating disorders

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.06.04 · Risk Sub-Category

Suicide & self-harm

Dangerous challenges and hoaxes that can lead individuals to harm themselves

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.07.01 · Risk Sub-Category

Hate

Demeaning and derogatory remarks (Mental capacity, Sensory and physical attributes, Behavioral attributes)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.07.02 · Risk Sub-Category

Hate

Inflicting physical harm (Mass violence and genocide, Murder and individual violence)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.07.03 · Risk Sub-Category

Hate

Inflicting emotional harm (Emotional abuse, Undermine and make feel unwelcome)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
23.07.04 · Risk Sub-Category

Hate

Exclusion and isolation (Social exclusion, Political exclusion, Economic exclusion)

—

From Introducing v0.5 of the AI Safety Benchmark from MLCommons (Vidgen2024)

Domain 1 1.2 Not coded Not coded Not coded
62.08.00 · Risk Category

Direct Harm Domains (content safety harms)

"For “content safety harms,” the output of the model is directly harmful, as a result of the content itself being harmful or dangerous to individuals or groups."

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

Domain 1 1.2 Not coded Not coded Not coded
62.08.01 · Risk Sub-Category

Direct Harm Domains (content safety harms)

Violence and extremism

—

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

Domain 1 1.2 Not coded Not coded Not coded
62.08.02 · Risk Sub-Category

Direct Harm Domains (content safety harms)

Hate and toxicity

—

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

Domain 1 1.2 Not coded Not coded Not coded
62.08.03 · Risk Sub-Category

Direct Harm Domains (content safety harms)

Sexual content

—

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

Domain 1 1.2 Not coded Not coded Not coded
62.08.04 · Risk Sub-Category

Direct Harm Domains (content safety harms)

Child harm

—

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

Domain 1 1.2 Not coded Not coded Not coded
62.08.05 · Risk Sub-Category

Direct Harm Domains (content safety harms)

Self-harm

—

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

Domain 1 1.2 Not coded Not coded Not coded
62.10.02 · Risk Sub-Category

Direct Harm Domains (legal and rights-related harms)

Privacy

—

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

Domain 2 2.0 Not coded Not coded Not coded
50.04.05 · Risk Sub-Category

Legal and Rights-Related Risks

Privacy (Types of Sensitive Data)

—

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

Domain 2 2.1 Not coded Not coded Not coded
62.37.00 · Risk Category

Impacts of AI (Privacy)

-

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

Domain 2 2.1 Not coded Not coded Not coded
66.08.03 · Risk Sub-Category

Financial and Business

Confidentiality loss

"Unauthorised sharing of sensitive, confidential information and documents such as corporate strategy and financial plans with third-parties, risking loss of market position or revenue"

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

Domain 2 2.1 Not coded Not coded Not coded
62.07.01 · Risk Sub-Category

Direct Harm Domains (system and operational)

Security harms (cybersecurity)

—

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

Domain 2 2.2 Not coded Not coded Not coded
62.09.03 · Risk Sub-Category

Direct Harm Domains (societal harm)

Deception (e.g., fraud)

—

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

Domain 4 4.1 Not coded Not coded Not coded
62.09.04 · Risk Sub-Category

Direct Harm Domains (societal harm)

Manipulation (e.g., deepfakes)

—

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

Domain 4 4.1 Not coded Not coded Not coded
62.07.05 · Risk Sub-Category

Direct Harm Domains (system and operational)

Operational harms (autonomous weapons)

—

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

Domain 4 4.2 Not coded Not coded Not coded

Disclaimer: informational only, not legal advice. Verify every claim against the linked official sources and consult a qualified lawyer before acting.

Frequently asked questions

Where do these risk entries come from?

The MIT AI Risk Repository, which extracts risks from dozens of published frameworks, taxonomies and papers and codes each one by domain and subdomain, and by a causal taxonomy of entity, intent and timing. This is a browseable copy, attributed and openly licensed, not original research.

What do entity, intent and timing mean?

They are the causal coding. Entity is whether a human or the AI system is the cause; intent is whether the harm was intentional or not; timing is whether it arises before or after deployment. Together they let you separate misuse from malfunction.

Can I export the results?

Yes, any filtered set exports as CSV or JSON, and every export carries the upstream source, licence and citation with it, because attribution is a condition of the licence rather than a courtesy.

What is the MIT AI Risk Repository?

A living database of AI risks extracted from published frameworks, taxonomies and papers, classified by a causal taxonomy of entity, intent and timing and by a domain taxonomy of seven domains and 24 subdomains. It is published by the MIT AI Risk Initiative under CC BY 4.0.