Public Database
Case Studies
Every approved AI failure case, classified against the AI Blindspot Framework. New to AIBlindspot? Start with the overview or the methodology.
Goal Misgeneralisation: AI Pursues Wrong Objectives After Deployment
Recent case. Full summary visible to registered users — sign in to read.
LLM Training Data Memorisation Enables Personal Data Extraction
Recent case. Full summary visible to registered users — sign in to read.
LLMs Can Link Personal Identifiers to Expose Private Individual Data
Recent case. Full summary visible to registered users — sign in to read.
AI Assistants Unable to Represent Core Ethical Concepts
Recent case. Full summary visible to registered users — sign in to read.
Model Bias Arising from Algorithm Design Choices Beyond Training Data
Recent case. Full summary visible to registered users — sign in to read.
Data Poisoning Attacks Corrupt Generative AI Training Datasets
Recent case. Full summary visible to registered users — sign in to read.
Training Data Poisoning Introduces Hidden Backdoors in Large Language Models
Recent case. Full summary visible to registered users — sign in to read.
AGI Goal Misalignment During Self-Improvement
Recent case. Full summary visible to registered users — sign in to read.
LLM Distributed Training Infrastructure Exposed to Network Disruption Attacks
Recent case. Full summary visible to registered users — sign in to read.
Training Data Poisoning and Backdoor Triggers in Large Language Models
Recent case. Full summary visible to registered users — sign in to read.
Noisy Training Data Causes LLM Hallucinations at Scale
Recent case. Full summary visible to registered users — sign in to read.
Deep Learning Framework Vulnerabilities Expose LLM Infrastructure
Recent case. Full summary visible to registered users — sign in to read.
Race Dynamics Drive Development of Unsafe Artificial General Intelligence
Recent case. Full summary visible to registered users — sign in to read.
Legal and Risk Frameworks Found Unequipped for Artificial General Intelligence
Recent case. Full summary visible to registered users — sign in to read.
Generative AI Disrupts Copyright Ownership and Authorship Norms
Recent case. Full summary visible to registered users — sign in to read.
Python Interpreter Vulnerabilities Expose LLM Infrastructure
Recent case. Full summary visible to registered users — sign in to read.
LLM Software Supply Chain Vulnerabilities Expose Development Pipelines
Recent case. Full summary visible to registered users — sign in to read.
LLM Development Toolchain Vulnerabilities Create Security Exposure
Recent case. Full summary visible to registered users — sign in to read.
GPU Side-Channel Attacks Enable Extraction of Trained LLM Parameters
Recent case. Full summary visible to registered users — sign in to read.
Toxic and Biased Training Data Embedded in Large Language Models
Recent case. Full summary visible to registered users — sign in to read.
Hardware Memory Attacks Enable Covert Manipulation of AI Model Parameters
Recent case. Full summary visible to registered users — sign in to read.
Toxic Training Data Corrupts LLM Output Quality and Safety
Recent case. Full summary visible to registered users — sign in to read.
LLM Decoding Randomness Causes Compounding Hallucination Errors
Recent case. Full summary visible to registered users — sign in to read.
Adversarial Input Manipulation Causes AI Model Prediction Failures
Recent case. Full summary visible to registered users — sign in to read.
Beyond accidental failureNational Security
We also track 20 hostile uses of AI.
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