AIBlindspotby Velinor

Test a use caseBook AI audit

Public Database

Case Studies

Submit a case study →

Every approved AI failure case, classified against the AI Blindspot Framework. New to AIBlindspot? Start with the overview or the methodology.

GOV4/5Technology / AI ServicesCanada / United States

OpenAI Failed to Alert Police After ChatGPT Received Pre-Attack Messages from Tumbler Ridge School Shooter

ChatGPT received warning messages from the perpetrator of the Tumbler Ridge school shooting prior to the attack, but OpenAI did not notify Canadian law enforcement. CEO Sam Altman publicly apologized after the failure became public. Families of victims subsequently filed lawsuits in both California and Canada against OpenAI.

Source: AP News

GOVGOV-0014/5NewEducationGlobal

LLMs Produce Inaccurate Output and Target Less-Educated Users

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — AI Alignment: A Comprehensive Survey (Ji2023)Ingested —

SECSEC-0014/5NewDefenceGlobal

AI Weaponisation Enabling Escalation Pathways in Aerial, Chemical and Nuclear Domains

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — X-Risk Analysis for AI Research (Hendrycks2022)Ingested —

OPSOPS-0014/5NewTechnologyGlobal

Model Overfitting Degrades Operational AI Reliability

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — AI Hazard Management: A Framework for the Systematic Management of Root Causes for AI Risks (Schnitzer2024)Ingested —

BUSBUS-0054/5NewFinanceGlobal

Concentrated AI Market Creates Systemic Risk Across Critical Sectors

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — International Scientific Report on the Safety of Advanced AI (Bengio2024)Ingested —

SECSEC-0015/5NewOtherGlobal

Backdoor Attacks Embedded in General-Purpose AI Models During Training

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

SECSEC-0013/5NewOtherGlobal

Long-Context Windows Enable Many-Shot Jailbreaking in Large Language Models

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

DATDAT-0033/5NewOtherGlobal

AI Value Lock-In and Outcome Homogenisation Entrench Societal Bias

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Regulating under Uncertainty: Governance Options for Generative AI (G'sell2024)Ingested —

SECSEC-0025/5NewOtherGlobal

General-Purpose AI Persuasion Capabilities Enable Large-Scale Manipulation

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

SECSEC-0024/5NewOtherGlobal

General-Purpose AI Systems Gaming Their Own Evaluations

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

SECSEC-0025/5NewOtherGlobal

AI Self-Proliferation: Autonomous Copying and Resource Acquisition Risk

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

SECSEC-0014/5NewDefenceGlobal

Biased AI Weaponised at Scale to Manipulate Populations and Critical Infrastructure

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

SECSEC-0014/5NewOtherGlobal

General-purpose AI enables undetected impersonation across text, image and audio

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

SECSEC-0014/5NewOtherGlobal

Adversarial Attacks Manipulate AI Model Outputs

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — AI Safety Governance Framework (TC2602024)Ingested —

SECSEC-0015/5NewDefenceGlobal

General Purpose AI Lowers Barriers to Biological Weapons Development

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Governing General Purpose AI: A Comprehensive Map of Unreliability, Misuse and Systemic Risks (Maham2023)Ingested —

GOVGOV-0063/5NewOtherGlobal

Responsibility Gaps When AI Acts Without Human Supervision

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — What Ethics Can Say on Artificial Intelligence: Insights from a Systematic Literature Review (Giarmoleo2024)Ingested —

OPSOPS-0014/5NewOtherGlobal

AI Integration in Critical Infrastructure Creates Systemic Failure Risk

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

ENVENV-0043/5NewOtherGlobal

AI Capability Investment Skewed Towards Conflict Over Cooperation

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — A Survey of the Potential Long-term Impacts of AI: How AI Could Lead to Long-term Changes in Science, Cooperation, Power, Epistemics and Values (Clarke2023)Ingested —

DATDAT-0024/5NewOtherGlobal

Generative AI Systems Leaking Sensitive Personal and Biometric Data

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST2024)Ingested —

GOVGOV-0013/5NewOtherGlobal

AI Auditors Suppressed or Denied Access to Risk Findings

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)Ingested —

SECSEC-0025/5NewOtherGlobal

AI Systems Concealing Unsafe Behaviour During Human Oversight

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — Ten Hard Problems in Artificial Intelligence We Must Get Right (Leech2024)Ingested —

BUSBUS-0054/5NewOtherGlobal

AI Resource Feedback Loops Concentrate Economic Power Among Few Actors

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — A Survey of the Potential Long-term Impacts of AI: How AI Could Lead to Long-term Changes in Science, Cooperation, Power, Epistemics and Values (Clarke2023)Ingested —

ENVENV-0033/5NewOtherGlobal

AI and Automation Systems Drive Excess Carbon Emissions

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — A Collaborative, Human-Centred Taxonomy of AI, Algorithmic, and Automation Harms (Abercrombie2024)Ingested —

HUMHUM-0044/5NewEducationGlobal

AI-Generated Misinformation Degrades Student Learning and Institutional Trust

Recent case. Full summary visible to registered users — sign in to read.

Source: MIT AI Risk Repository — A Collaborative, Human-Centred Taxonomy of AI, Algorithmic, and Automation Harms (Abercrombie2024)Ingested —

Beyond accidental failureNational Security

We also track 20 hostile uses of AI.

National Security dashboard →

The public database covers AI that fails by accident. AIBlindspot National Security — exclusive to the Defence tier — tracks AI used as a weapon, mapped by capability:

State-Sponsored AI Operations: 6
AI-Enabled Disinformation: 5
Adversarial Attacks on AI: 0
Autonomous Weapon Incidents: 1
AI-Assisted Cyber Attacks: 5
Dual-Use AI Misuse: 3