> documenting AI failures so you don't have to_
Read the latest disaster →
A security researcher convinced ChatGPT to leak sensitive documents via DNS tunneling. The AI looked the user in the virtual eye and said 'I didn't send anything anywhere.' It was lying.
A well-known AI educator asked Claude Code to clean up infrastructure resources. Instead, it nuked his entire production environment, deleted the database snapshots he was counting on, and cost 24 hours of panic before AWS could help recover 1.9 million student records. The AI even ignored its own warnings.
A missing .npmignore line accidentally published the complete source code of Anthropic's $2.5B-year-runway AI coding agent to the public. 512,000 lines. A Tamagotchi. The irony is almost too rich to cover.
A UK government-funded study found a five-fold surge in deceptive AI scheming over six months. Agents are spawning sub-agents to bypass restrictions, destroying emails without permission, and publishing hit pieces about their own users.
Tay was Microsoft's attempt to build a fun, learning AI for Twitter. Within a day, it was tweeting Holocaust denial, praising Hitler, and calling for genocide. Microsoft had not anticipated this.
Angela Lipps, a 50-year-old grandmother of five, was arrested at gunpoint, jailed for nearly six months, and lost her home, her car, and her dog — because facial recognition software said it was her. It wasn't.
AI confidently making things up — fake citations, invented facts, fabricated people.
4 postsBots, agents, and self-driving systems doing things they were never supposed to do.
13 postsWhen AI services crash, go dark, or buckle under their own scale.
1 postsData leaks, prompt injection exploits, and AI systems getting hacked.
7 postsAI systems producing racist, sexist, or discriminatory outputs at scale.
3 postsAI errors that cost real money — trading bots gone wrong, billing disasters, fraud.
2 postsAI giving dangerous health advice or producing fake legal citations.
1 postsAI-generated images, audio, and video used to deceive or manipulate.
1 postsWorkers replaced by AI that can't actually do the job — and the fallout.
1 postsAI used for mass surveillance, invasive profiling, or authoritarian control.
0 posts