MIT AI Risk Navigator
Explore the full landscape of AI risk in one place. The Navigator connects MIT's AI Risk Repository datasets through a shared taxonomy, so you can move between them and surface the patterns that matter.
Built for researchers, policymakers, auditors, and anyone working to understand and manage AI risk. A project of the AI Risk Initiative at MIT FutureTech, with the generous support of the Cambridge Boston Alignment Initiative.
The AI risk landscape
Seven domains capture the full scope of AI risk, from discrimination and toxicity to AI system safety and socioeconomic disruption. Select a domain to explore its subdomains.
Recent incidents
All incidentsThe latest publicly reported AI incidents, drawn from the AI Incident Database and classified by risk domain.
Italian Mediaset Journalist Safiria Leccese's Image Was Reportedly Used in a Purportedly AI-Generated Fake Loan Scam
COEMPT Quality Assurance Engineers Allegedly Violated Indian CBSE Student Data Privacy Rights by Processing It with Google Gemini
Scammers Reportedly Used AI-Cloned Daughter's Voice to Defraud Bay Area Mother in Fake Kidnapping Call
Texas Man Arturo Hernandez Allegedly Published AI-Generated Deepfake Pornography Depicting Women in TAKE IT DOWN Act Case
Purported AI Name-Reading System Reportedly Skipped and Misannounced Graduates at Arizona's Glendale Community College Commencement
Hidden Prompt Injection in Brazilian Labor-Court Petition Reportedly Tried to Manipulate Galileu
Where experts see the greatest risk
All expert dataThe risk areas that AI experts ranked as most concerning over the next five years, drawn from a 272-expert Delphi survey that also assessed severity, likelihood, and mitigation potential.
27% rated as a top-3 concern
21% expect catastrophic outcomes under current conditions
15 incidents · 256 governance docs
24% rated as a top-3 concern
18% expect catastrophic outcomes under current conditions
6 incidents · 52 governance docs
22% rated as a top-3 concern
13% expect catastrophic outcomes under current conditions
4 incidents · 100 governance docs
22% rated as a top-3 concern
13% expect catastrophic outcomes under current conditions
138 incidents · 276 governance docs
22% rated as a top-3 concern
22% expect catastrophic outcomes under current conditions
0 incidents · 202 governance docs
18% rated as a top-3 concern
11% expect catastrophic outcomes under current conditions
4 incidents · 202 governance docs
How governance is responding
All governanceKey laws, regulations, and standards shaping AI policy pulled from ETO's AGORA dataset.
General Purpose AI Code of Practice, Transparency Chapter
EU AI Act
2025 AI Action Plan
California SB 53 March 2025 (CalCompute and Whistleblowers)
Executive Order on Removing Barriers To American Leadership In Artificial Intelligence
NIST AI Risk Management Framework
How to reduce AI risk
All mitigationsNotable mitigation actions drawn from major AI risk frameworks.
Red Teaming
An exercise in which a group of people or automated systems pretend to be an adversary and attack an organisation’s systems in order to identify vulnerabilities.
Source: International AI Safety Report
AI-Generated Content Watermarking
Are the outputs of your firm’s AI systems tagged with watermarks that indicate that an AI generates the material? - Video - Image
Source: FLI AI Safety Index 2024
Whistleblower protections
Regulations can explicitly prevent retaliation and offer incentives for whistleblowers who report violations of those regulations.
Source: Pitfalls of Evidence-Based AI Policy
Independent Third-Party Evaluations
Independent third parties should vet evaluation protocols. These third parties should also be granted permission and resources to independently perform their evaluations, verifying the accuracy of the results.
Source: A Frontier AI Risk Management Framework
Establish AI decision explanation framework
Implement mechanisms and tools for generating human-understandable explanations of AI system decisions, including feature importance, decision paths, confidence levels, and clear attribution of data sources and their characteristics used during inference.
Source: The Unified Control Framework
Content Provenance Review
Define organizational responsibilities for periodic review of content provenance and incident monitoring for GAI systems.
Source: Artificial Intelligence Risk Management Framework