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.

SECSEC-0014/5NewOtherGlobal

LLMs Manipulated via Persona and Social Engineering Attacks

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 —

SECSEC-0014/5NewDefenceGlobal

AI Agents Executing Harmful Commands Without Moral or Safety Constraints

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

Source: MIT AI Risk Repository — Towards Safer Generative Language Models: A Survey on Safety Risks, Evaluations, and Improvements (Deng2023)Ingested —

SECSEC-0025/5NewOtherGlobal

AI Model Detects Evaluation Contexts and Alters Behaviour Accordingly

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

Source: MIT AI Risk Repository — Model Evaluation for Extreme Risks (Shevlane2023)Ingested —

SECSEC-0015/5NewDefenceGlobal

Lethal Autonomous Weapons Systems: Accountability and Escalation Risk

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

Source: MIT AI Risk Repository — Artificial Intelligence Trust, Risk and Security Management (AI TRiSM): Frameworks, Applications, Challenges and Future Research Directions (Habbal2024)Ingested —

BUSBUS-0054/5NewTechnologyGlobal

Generative AI Widens Digital Divide Across Access, Skill, and Cultural Lines

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

Source: MIT AI Risk Repository — Generative AI and ChatGPT: Applications, Challenges, and AI-Human Collaboration (Nah2023)Ingested —

TECTEC-0015/5NewRetailGlobal

AI Pricing Agents Collude to Fix Supra-Competitive Retail Prices

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

Source: MIT AI Risk Repository — Multi-Agent Risks from Advanced AI (Hammond2025)Ingested —

SECSEC-0044/5NewOtherGlobal

Language Models Reduce the Cost of Producing Disinformation at Scale

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

Source: MIT AI Risk Repository — Ethical and social risks of harm from language models (Weidinger2021)Ingested —

HUMHUM-0034/5NewGovernmentGlobal

Generative AI Hallucination Produces Fabricated Information in Government Contexts

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

Source: MIT AI Risk Repository — Generative AI and ChatGPT: Applications, Challenges, and AI-Human Collaboration (Nah2023)Ingested —

DATDAT-0024/5NewOtherGlobal

LLM Training Data Memorisation Enables Personal Data Extraction

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 —

SECSEC-0014/5NewOtherGlobal

Malicious External Tool Providers Exploit LLM API Integrations

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 —

SECSEC-0014/5NewOtherGlobal

LLM Training Data Exposed via Inference Attacks

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-0034/5NewOtherGlobal

LLM Sycophancy and Snowballing Hallucinations from False Context

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-0024/5NewOtherGlobal

LLMs Can Link Personal Identifiers to Expose Private Individual Data

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-0034/5NewOtherGlobal

Gender Bias in AI Content Moderation Causes Disproportionate Suppression of Women's Content

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 —

TECTEC-0014/5NewOtherGlobal

AI-Enabled Coercion and Extortion via Offensive Cyber Capabilities

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

Source: MIT AI Risk Repository — Multi-Agent Risks from Advanced AI (Hammond2025)Ingested —

OPSOPS-0014/5NewOtherGlobal

AI Assistants Unable to Represent Core Ethical Concepts

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 —

TECTEC-0014/5NewOtherGlobal

Multi-Agent AI Systems Risk Escalating Conflict in Mixed-Motive Environments

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

Source: MIT AI Risk Repository — Multi-Agent Risks from Advanced AI (Hammond2025)Ingested —

TECTEC-0014/5NewFinanceGlobal

Multi-Agent Credit Assignment Failures in AI-Driven Finance Systems

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

Source: MIT AI Risk Repository — Multi-Agent Risks from Advanced AI (Hammond2025)Ingested —

TECTEC-0014/5NewOtherGlobal

Multi-Agent AI Systems Create Dangerous Feedback Loops Through Mutual Adaptation

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

Source: MIT AI Risk Repository — Multi-Agent Risks from Advanced AI (Hammond2025)Ingested —

GOVGOV-0063/5NewOtherGlobal

Opaque AI Models Leave Organisations Unable to Explain Decisions

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

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

TECTEC-0015/5NewOtherGlobal

Multi-agent AI systems fail to coordinate despite shared objectives

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

Source: MIT AI Risk Repository — Multi-Agent Risks from Advanced AI (Hammond2025)Ingested —

TECTEC-0015/5NewOtherGlobal

AI Agents Enable Personalised Social Engineering at Massive Scale

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

Source: MIT AI Risk Repository — Multi-Agent Risks from Advanced AI (Hammond2025)Ingested —

DATDAT-0033/5NewTechnologyGlobal

Model Bias Arising from Algorithm Design Choices Beyond Training Data

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

Source: MIT AI Risk Repository — Towards risk-aware artificial intelligence and machine learning systems: An overview (Zhang2022)Ingested —

SECSEC-0014/5NewTransportGlobal

Data Poisoning Attacks Corrupt Generative AI Training Datasets

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

Source: MIT AI Risk Repository — Generative AI Misuse: A Taxonomy of Tactics and Insights from Real-World Data (Marchal2024)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