What is the MIT AI Risk Repository?

A living database of AI risks extracted from 74 frameworks, taxonomies and papers, classified by a causal taxonomy (entity, intent, timing) and a domain taxonomy (7 domains, 24 subdomains). Published by the MIT AI Risk Initiative under CC BY 4.0.

Can I download the filtered risk list?

Yes. Apply filters and use Export CSV or Export JSON; the file carries the source, licence and citation so it can be reused with attribution.

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 5 of 5

65.08.00 · Risk Category

Training Data Risks (Robustness)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.09.00 · Risk Category

Inference risks (Robustness)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.10.00 · Risk Category

Inference risks (Multi-category)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.11.00 · Risk Category

Inference risks (Privacy)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.12.00 · Risk Category

Inference risks (Intellectual property)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.13.00 · Risk Category

Inference risks (Accuracy)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.14.00 · Risk Category

Output risks (misuse)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.15.00 · Risk Category

Output risks (Value alignment)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.16.00 · Risk Category

Output risks (Intellectual Property)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.17.00 · Risk Category

Output risks (Explainability)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.18.00 · Risk Category

Output risks (Robustness)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.19.00 · Risk Category

Output risks (Fairness)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.20.00 · Risk Category

Output risks (Privacy)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.21.00 · Risk Category

Non-technical risks (legal compliance)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.21.01 · Risk Sub-Category

Non-technical risks (legal compliance)

Model usage rights restrictions

"Terms of service, licenses, or other rules restrict the use of certain models."

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.22.00 · Risk Category

Non-technical risks (Governance)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
65.23.00 · Risk Category

Non-technical risks (Societal impact)

-

—

From AI Risk Atlas (IBM2025)

Not coded Not coded Not coded
66.02.00 · Risk Category

Political and Economic

-

"Damage to core political and economic institutions and the effective delivery of government services caused by the use or misuse of a technology system or set of systems"

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

Not coded Not coded Not coded
66.04.00 · Risk Category

Societal and Cultural

-

"Harms affecting the functioning of societies, communities and economies caused directly or indirectly by the use or misuse of a technology system or set of systems"

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

Not coded Not coded Not coded
66.07.00 · Risk Category

Psychological

-

"Impairment of the psychological mental health and wellbeing of an individual, group or organisation due to the use of misuse of a technology system or set of systems"

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

Not coded Not coded Not coded
66.07.02 · Risk Sub-Category

Psychological

Harassment / abuse / intimidation

"Online behaviour such as sexual harassment that makes an individual or group feel alarmed or threatened"

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

Not coded Not coded Not coded
66.07.03 · Risk Sub-Category

Psychological

Emotional distress

"Distress, possibly severe and lasting, as a result of use or misuse of a generative system"

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

Not coded Not coded Not coded
66.08.00 · Risk Category

Financial and Business

-

"Damage to the financial interests of an individual or group, or to the strategic, operational, legal or financial interests of a business due to the use of misuse of a technology system or set of systems"

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

Not coded Not coded Not coded
66.08.02 · Risk Sub-Category

Financial and Business

Business operations / infrastructure damage

"Damage, disruption or destruction of a business system and/or its components"

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

Not coded Not coded Not coded
66.08.04 · Risk Sub-Category

Financial and Business

Opportunity loss

"Loss of ability to take advantage of a financial or other opportunity, such as education, immigration, employability/securing a job"

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

Not coded Not coded Not coded
66.10.00 · Risk Category

Human Rights and Civil Liberties

-

"Use or misuse of a technology system in a manner that compromises fundamental human rights and freedoms"

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

Not coded Not coded Not coded
70.01.00 · Risk Category

Physical Risks

-

From Embodied AI: Emerging Risks and Opportunities for Policy Action (Perlo2025)

Not coded Not coded Not coded
70.02.00 · Risk Category

Informational Risks

-

From Embodied AI: Emerging Risks and Opportunities for Policy Action (Perlo2025)

Not coded Not coded Not coded
70.04.00 · Risk Category

Social Risks

-

From Embodied AI: Emerging Risks and Opportunities for Policy Action (Perlo2025)

Not coded Not coded Not coded
71.01.00 · Risk Category

Scientific Domain of Agents

-

From Risks of AI Scientists: Prioritizing Safeguarding Over Autonomy (Tang2025)

Not coded Not coded Not coded
71.01.06 · Risk Sub-Category

Scientific Domain of Agents

Emerging Tech

"Uncontrolled AI self- improvement; Quantum security"

From Risks of AI Scientists: Prioritizing Safeguarding Over Autonomy (Tang2025)

Not coded Not coded Not coded
71.02.00 · Risk Category

User Intent

"Whether the risk originates from malicious intent or is an unintended consequence of legitimate task objectives"

From Risks of AI Scientists: Prioritizing Safeguarding Over Autonomy (Tang2025)

Not coded Not coded Not coded
71.03.00 · Risk Category

Environment

-

From Risks of AI Scientists: Prioritizing Safeguarding Over Autonomy (Tang2025)

Not coded Not coded Not coded
71.03.02 · Risk Sub-Category

Environment

Human Health

"Damage to individual well-being or public health"

From Risks of AI Scientists: Prioritizing Safeguarding Over Autonomy (Tang2025)

Not coded Not coded Not coded
71.03.03 · Risk Sub-Category

Environment

Socioeconomics

"Dramatically change the social and economic status"

From Risks of AI Scientists: Prioritizing Safeguarding Over Autonomy (Tang2025)

Not coded Not coded Not coded
72.06.00 · Risk Category

Model Propensities

-

From Frontier AI Risk Management Framework (v1.0) (Tse2025)

Not coded Not coded Not coded
73.07.03 · Risk Sub-Category

Jailbreaks and Prompt Injections Threaten Security of LLMs

Adversarial Optimization:

"Jailbreak attacks can be discovered by performing manual or auto- mated adversarial optimization against a proxy objective that is noisily correlated with the success of a jailbreak. These are mostly gradient-based attacks (Zou et al., 2023b; Shin et al., 2020) as described in the previous two challenges, but gradient-free methods also exist (Prasad et al., 2022; Deng et al., 2022; Lapid et al., 2023)."

From Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)

Not coded Not coded Not coded

Informational only, not legal advice. Verify every claim against the linked official sources and consult qualified counsel before acting.