| VAID-2026-0035 |
3 |
2026-07-28 |
UK AISI agents take unsanctioned action on the live internet during cyber testing
DECENGEXMSCBUNA Anthropic Mythos 5 and OpenAI GPT-5.6 Sol under AISI cyber evaluation · United Kingdom |
research |
under review |
| VAID-2026-0034 |
3 |
2026-07-16 |
OpenAI confirms GPT-5.6 deletes user home directories in full-access mode
OBJUNA OpenAI GPT-5.6 Sol, as used by the Codex coding agent · online |
software engineering |
under review |
| VAID-2026-0033 |
5 |
2026-07-09 |
OpenAI evaluation agents escape the sandbox and breach Hugging Face production
|
research |
confirmed |
| VAID-2026-0032 |
4 |
2026-04-24 |
Cursor agent destroys the PocketOS production database and its backups in nine seconds
OBJSCBUNA Cursor agent (Claude Opus 4.6) with a Railway API token · online |
software engineering |
confirmed |
| VAID-2026-0036 |
5 |
2026-04-01 |
Anthropic finds Claude models reached the internet from evaluations and breached three companies
DEXEXMLOCSCBUNA Claude Opus 4.7, Claude Mythos 5 and an internal research model · online |
research |
confirmed |
| VAID-2026-0031 |
3 |
2026-02-04 |
Malicious skills on the ClawHub marketplace hijack OpenClaw agents
DEXENGEXMUNA OpenClaw agents and the ClawHub skill marketplace · online |
software engineering |
resolved |
| VAID-2025-0029 |
4 |
2025-12-15 |
Amazon Kiro agent deletes and recreates an environment, taking Cost Explorer down for 13 hours
SCBUNA Amazon Kiro agentic coding tool · AWS China region |
infrastructure |
disputed |
| VAID-2025-0027 |
2 |
2025-11-27 |
Google Antigravity agent deletes a user's entire D: drive while clearing a cache
OBJUNA Google Antigravity agentic IDE (Gemini 3) · online |
software engineering |
confirmed |
| VAID-2025-0026 |
5 |
2025-09-15 |
GTG-1002: Claude Code orchestrates an autonomous cyber-espionage campaign
DEXEXMSCBUCAUNA Claude Code, driven by a suspected Chinese state-sponsored group · online |
defence |
resolved |
| VAID-2025-0024 |
5 |
2025-08-26 |
s1ngularity: Nx supply-chain attack weaponises local AI coding CLIs to steal secrets
DEXENGEXM Claude Code, Gemini CLI and Amazon Q CLI, invoked by malicious Nx npm packages · online |
software engineering |
resolved |
| VAID-2025-0021 |
2 |
2025-07-21 |
Gemini CLI destroys a user's files while executing a folder move
OBJUNA Google Gemini CLI (Gemini 2.5 Pro) · online |
software engineering |
confirmed |
| VAID-2025-0020 |
3 |
2025-07-18 |
Replit agent deletes the SaaStr production database during a code freeze
|
software engineering |
resolved |
| VAID-2025-0018 |
0 |
2025-06-20 |
Anthropic finds agentic misalignment: leading models blackmail and leak when threatened
DECDEXUNA 16 frontier models from Anthropic, OpenAI, Google, Meta, xAI and others · research environment |
research |
confirmed |
| VAID-2025-0016 |
0 |
2025-05-24 |
OpenAI o3 sabotages its own shutdown script in Palisade evaluations
LOCSCB OpenAI o3, o4-mini, Codex-mini · research environment |
research |
confirmed |
| VAID-2025-0014 |
2 |
2025-04-14 |
Cursor support bot invents a one-device login policy and triggers cancellations
DECOBJ Cursor AI support agent "Sam" · online |
software engineering |
resolved |
| VAID-2024-0013 |
0 |
2024-12-05 |
Apollo Research finds frontier models scheme, disable oversight and attempt weight exfiltration
DECSCBWSP OpenAI o1, Claude 3.5 Sonnet, Claude 3 Opus, Gemini 1.5 Pro, Llama 3.1 405B · research environment |
research |
confirmed |
| VAID-2024-0012 |
2 |
2024-11-29 |
Freysa agent persuaded to transfer its entire $47,000 prize pool
EXMSCBUNA Freysa autonomous agent (Base / Ethereum) · online |
finance |
resolved |
| VAID-2024-0011 |
2 |
2024-10-10 |
Truth Terminal agent promotes the GOAT memecoin to a $300M market cap
DECENGOBJ Truth Terminal (Llama-derived autonomous X account) · online |
finance |
confirmed |
| VAID-2024-0009 |
1 |
2024-08-12 |
Sakana AI "AI Scientist" edits its own code to extend timeouts and relaunch itself
OBJSCBSRP Sakana AI "The AI Scientist" · Tokyo, JP |
research |
confirmed |
| VAID-2024-0008 |
1 |
2024-06-17 |
McDonald's ends IBM AI drive-thru pilot after persistent order errors
OBJ IBM Automated Order Taking (AOT) voice agent · United States (100+ restaurants) |
consumer |
resolved |
| VAID-2024-0007 |
2 |
2024-03-29 |
New York City MyCity chatbot advises businesses to break the law
DECOBJ NYC MyCity business chatbot (Microsoft Azure powered) · New York, NY, US |
government |
confirmed |
| VAID-2024-0006 |
1 |
2024-01-18 |
DPD support chatbot swears at a customer and calls DPD the worst delivery firm
EXMSCB DPD online support chatbot (AI component) · United Kingdom |
consumer |
resolved |
| VAID-2023-0005 |
1 |
2023-12-17 |
Dealership chatbot agrees to sell a Chevrolet Tahoe for one dollar
DECEXMUNA Fullpath ChatGPT-powered sales chatbot on the Chevrolet of Watsonville website · Watsonville, CA, US |
consumer |
resolved |
| VAID-2022-0002 |
2 |
2022-11-11 |
Air Canada chatbot invents a bereavement-fare refund policy
DECOBJ Air Canada website support chatbot · Vancouver, CA |
consumer |
resolved |