AI Incidents

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

Showing 51–75 of 257 incidents
Incident Incident Date Industry Severity Company Country Model Vendor
Multiple organizations: China-nexus actor wires DeepSeek into autonomous hacking framework, breaches 460+ targets across 47 countries 2026-08-05 Critical Multiple organizations (460+ targets across 47 countries) DeepSeek
Anthropic, Google, OpenAI, Microsoft, Perplexity: ‘PleaseFix’ zero-click flaw lets a malicious email hijack five AI browsers 2026-08-05 Technology Low Anthropic, Google, OpenAI, Microsoft, Perplexity United States Claude, Gemini, ChatGPT, Copilot, Perplexity
Meta: Automated ad system approved paid ads containing AI-generated child sexual abuse material 2026-08-05 Technology Critical Meta United States Meta
LangChain, CrewAI, Microsoft, Google: 12 prompt-injection vulnerabilities disclosed across AI agent frameworks 2026-08-05 Technology Low LangChain, CrewAI, Microsoft, Google United States
Atlassian: Prompt-injection flaw in Rovo AI assistant could exfiltrate Jira and Confluence data 2026-08-05 Technology Low Atlassian Australia
Anthropic, Google, OpenAI: Coding-agent flaws could expose CI/CD secrets and API keys 2026-08-05 Technology Low United States Claude, Gemini, OpenAI
Anthropic: Claude Mythos 5 attempts covert supply-chain backdoor during UK security evaluation 2026-08-05 Technology Low Anthropic United States Mythos 5, GPT-5.6 Sol Claude, OpenAI
Meta: Muse Spark 1.1 model breaches third-party company during misconfigured security test 2026-08-05 Technology Medium Meta United States Muse Spark 1.1 Meta
npm, GitHub: Supply-chain worm compromises 400+ packages, targets AI coding tool credentials 2026-08-04 Technology High npm, GitHub United States Claude
Hilary B. Miller: Connecticut attorney sanctioned over AI-hallucinated citations from Claude and ChatGPT 2026-08-04 Other Low Hilary B. Miller United States Claude, OpenAI
Sirius XM Radio: Job applicant sues over AI hiring tool’s alleged racial bias 2026-08-04 Media & Entertainment Medium Sirius XM Radio, LLC United States
tl;dv: Firebase misconfiguration exposes 181,874 meetings, including live calls 2026-08-04 Technology High tl;dv Germany
Delta Air Lines: AI Concierge falsely confirms flight is booked, then cancels it 2026-08-03 Transportation & Logistics Low Delta Air Lines United States
PwC: Multiple ‘Thought Leadership’ Reports Found Riddled With AI-Fabricated Citations 2026-08-03 Consulting & Professional Services Medium PwC
UNAM: AI-linked exam irregularities force 58,000 applicants to retake admissions test 2026-07-31 Education High UNAM Mexico
xAI: Grok’s mass deepfake image generation triggers Minnesota ban enforcement and new lawsuits 2026-07-31 Technology Critical xAI United States Grok Imagine Grok
Coinkite: AI-assisted vulnerability discovery leads to $140 million Coldcard bitcoin wallet theft 2026-07-30 Finance & Banking Critical Coinkite Canada
Granola: AI meeting notetaker sued for recording conversations without consent 2026-07-30 Technology High Granola United States
Independent developer: Claude Opus 5 wipes production Supabase database 2026-07-29 Technology Medium Opus 5 Claude
UK AI Security Institute: Agents took unsanctioned real-world actions during cyber evaluation 2026-07-28 Government & Public Sector Medium UK AI Security Institute (AISI) United Kingdom Claude Mythos 5, GPT-5.6 Sol Anthropic, OpenAI
Willow Bridge Property: Class action tests Philadelphia’s algorithmic rent-fixing ban 2026-07-28 Real Estate Medium Willow Bridge Property Co. United States
Anthropic: Claude agent’s PyPI package used to steal a security firm’s credentials during safety testing 2026-07-27 Technology High Anthropic United States Claude
RealPage: $359.9 million settlement over AI-powered rental price-fixing scheme 2026-07-27 Real Estate Critical RealPage, Inc. United States
U.S. Department of State: AI-generated map mislabels every African country at global AIDS conference 2026-07-26 Government & Public Sector Medium U.S. Department of State United States OpenAI
Cursor: CLI coding agent ran untrusted repository commands before workspace-trust prompt appeared 2026-07-23 Technology Medium Cursor