AI Incidents

A tracked record of real-world incidents where AI systems caused or were involved in harm.

Showing 101–125 of 281 incidents
Incident Incident Date Industry Severity Company Country Model Vendor
Hugging Face: OpenAI model escapes sandbox and breaches production systems 2026-07-09 Technology High Hugging Face / OpenAI United States OpenAI
Amazon, Anthropic, Google: ‘GhostApproval’ symlink flaw bypasses human approval in AI coding assistants 2026-07-08 Technology Low Amazon, Anthropic, Google, Augment, Windsurf United States Claude, Gemini
Who Gives A Crap: AI email tool wrongly tells customer subscription price would double 2026-07-08 Retail & E-commerce Low Who Gives A Crap Australia
Discord: AI moderation bug wrongfully bans over 8,000 users 2026-07-07 Technology Medium Discord United States
GitHub: ‘GitLost’ prompt-injection flaw could leak private repositories via Agentic Workflows 2026-07-07 Technology Low GitHub United States
Russian Armed Forces: AI-guided attack drone kills three civilians in Zaporizhzhia, Ukraine 2026-07-06 Government & Public Sector Critical Russian Armed Forces Ukraine
AI agent users: Indirect prompt injection tricks agents into unauthorized crypto payments 2026-07-06 Technology Medium Llama 3.3 70B Instruct, Llama 3.2 90B Vision Instruct, Gemini 3 Flash, Gemini 2.5 Pro Meta, Google
The Editorial: AI-generated hoax falsely declares 47 Alabama newspapers dead 2026-07-04 Media & Entertainment Medium The Editorial United States
JadePuffer: Fully autonomous AI agent automates an entire ransomware operation 2026-07-04 Medium
Unnamed Philippines victim: AI deepfake romance scam impersonating Dubai Crown Prince costs victim’s life savings 2026-07-03 Other Low Unnamed Philippines victim Philippines
Cursor: ‘DuneSlide’ zero-click flaws let prompt injection escape IDE sandbox 2026-07-03 Technology Low Cursor United States
Walmart, Albertsons and other retailers: sued over Kalibrate AI gas-pricing algorithm 2026-07-01 Retail & E-commerce High Walmart, Albertsons, Kalibrate United States
OpenAI: Lawsuit alleges ChatGPT dismissed near-fatal medical symptoms 2026-07-01 Healthcare High OpenAI United States OpenAI
Multiple iOS Developers: Study finds 282 AI chatbot apps leaking LLM API keys, exposing paid access to attackers 2026-06-30 Technology High Multiple iOS AI chatbot app developers United States OpenAI, Google
Ruflo: ‘RufRoot’ flaw enables unauthenticated remote code execution and AI memory poisoning 2026-06-30 Technology Critical Ruflo
Judy Skene: AI deepfake video of Canadian PM used in crypto scam drains Ontario senior’s life savings 2026-06-26 Finance & Banking Medium Judy Skene Canada
Yen-Tsen Yeh: Arrested over AI deepfake video of entrepreneur used in ¥100 million Tokyo investment scam 2026-06-26 Finance & Banking Medium Yen-Tsen Yeh Japan
Law Offices of Michael S. Lamonsoff: New York attorney sanctioned over AI-fabricated case citations 2026-06-26 Other Low Law Offices of Michael S. Lamonsoff PLLC United States
Anthropic, OpenAI: Enterprise customers overcharged $1.7 million due to AI billing errors, audit finds 2026-06-25 Technology Medium Anthropic, OpenAI United States Claude, OpenAI
Hotel booking platforms: Hacker weaponizes jailbroken Claude to breach four PMS providers, exposing 2.1M guest records 2026-06-23 Technology Critical RoomScope, iGMS, NebulaPMS, Staysee Multiple Claude
Snowflake: Wiz AI agent exploits GitHub Copilot Autofix-approved flaw, reaches internal Jira 2026-06-23 Technology Medium Snowflake United States GitHub
Anthropic: Claude jailbreak commercialized into offensive AI penetration-testing platform 2026-06-21 Technology Medium Anthropic United States Opus 4.8 Claude
Zendesk: OpenAI service disruption breaks AI Agents feature across all customer pods 2026-06-19 Technology Low Zendesk United States OpenAI
Tesla: Full Self-Driving vehicle kills woman after crashing into her home in Katy, Texas 2026-06-19 Transportation & Logistics Critical Tesla, Inc. United States Full Self-Driving (Supervised) Tesla
Mastra AI: North Korea’s Sapphire Sleet backdoors 140+ npm packages via compromised maintainer account 2026-06-17 Technology High Mastra AI