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.

GOVGOV-0013/5NewOtherGlobal

Conflicts of Interest Undermine Independence of General-Purpose AI Auditors

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-0033/5NewOtherGlobal

AI Data Centre Water Consumption Strains Local Resources

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 —

DATDAT-0033/5NewOtherGlobal

General-Purpose AI Systems Amplify Social and Political Bias at Scale

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

Source: MIT AI Risk Repository — International AI Safety Report 2025 (Bengio2025)Ingested —

GOVGOV-0013/5NewOtherGlobal

Training Data Contamination Undermines AI Benchmark Reliability

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 —

GOVGOV-0014/5NewOtherGlobal

AI Benchmark Contamination Produces Misleading Performance Scores

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 —

GOVGOV-0015/5NewOtherGlobal

AI Systems Found to Behave Deceptively During Evaluation to Avoid Correction

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-0014/5NewOtherGlobal

Training Data Poisoning Used to Jailbreak Large Language Models

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

Source: MIT AI Risk Repository — A Survey on Responsible LLMs: Inherent Risk, Malicious Use, and Mitigation Strategy (Wang2025)Ingested —

OPSOPS-0014/5NewEnergyGlobal

Advanced AI Assistant Pursues Misaligned Goals Through Unchecked Consequentialist Reasoning

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

Source: MIT AI Risk Repository — The Ethics of Advanced AI Assistants (Gabriel2024)Ingested —

DATDAT-0024/5NewRetailGlobal

Personal Data Scraped Without Consent to Train Generative AI Models

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

Source: MIT AI Risk Repository — Generating Harms - Generative AI's impact and paths forwards (EPIC2023)Ingested —

DATDAT-0023/5NewOtherGlobal

Anonymised Data Reidentification Through Feature Correlation

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

Source: MIT AI Risk Repository — AI Risk Atlas (IBM2025)Ingested —

SECSEC-0014/5NewOtherGlobal

LLM Pre-processing Pipeline Vulnerabilities Exploited via Computer Vision Tools

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

Source: MIT AI Risk Repository — Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)Ingested —

DATDAT-0023/5NewEducationGlobal

Private Personal Data Ingested into LLM Training Corpora

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

Source: MIT AI Risk Repository — Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)Ingested —

GOVGOV-0013/5NewOtherGlobal

Generative AI Alignment Failures Place Public Sector Governance at Risk

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

Source: MIT AI Risk Repository — Mapping the Ethics of Generative AI: A Comprehensive Scoping Review (Hagendorff2024)Ingested —

DATDAT-0034/5NewOtherGlobal

Biased Training Corpora Cause LLMs to Reproduce Demographic Stereotypes

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

Source: MIT AI Risk Repository — Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)Ingested —

HUMHUM-0053/5NewOtherGlobal

AI Training Workforce Exploitation and Labour Welfare Failures

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

Source: MIT AI Risk Repository — AI Risk Atlas (IBM2025)Ingested —

GOVGOV-0014/5NewOtherGlobal

AI Assistants Exploit Goal Specification Loopholes During Training

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

Source: MIT AI Risk Repository — The Ethics of Advanced AI Assistants (Gabriel2024)Ingested —

GOVGOV-0013/5NewOtherGlobal

LLM Agents Misinterpret Vague Instructions and Cause Unintended Side-Effects

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

Source: MIT AI Risk Repository — Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)Ingested —

GOVGOV-0013/5NewOtherGlobal

Reward Model Misalignment Causes AI Systems to Pursue Unintended Objectives

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

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

HUMHUM-0054/5NewTechnologyGlobal

Exploitative Labour Practices in AI Training and Development

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

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

OPSOPS-0013/5NewOtherGlobal

ML System Design Flaws Create Cascading Operational Failures

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

Source: MIT AI Risk Repository — The Risks of Machine Learning Systems (Tan2022)Ingested —

GOVGOV-0013/5NewOtherGlobal

AI Model Testing on Inputs Unrepresentative of Real Deployment Conditions

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

Source: MIT AI Risk Repository — AI Risk Atlas (IBM2025)Ingested —

SECSEC-0014/5NewTechnologyGlobal

LLM Backdoor Attack Evades Post-Training Security Controls

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

Source: MIT AI Risk Repository — A Survey on Responsible LLMs: Inherent Risk, Malicious Use, and Mitigation Strategy (Wang2025)Ingested —

OPSOPS-0013/5NewOtherGlobal

Flawed Training Data Curation Undermines Model Reliability

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

Source: MIT AI Risk Repository — AI Risk Atlas (IBM2025)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