ssh: AI incidents
What has gone wrong with AI systems, where, and how badly.
Incidents, Oct 2026
Average of the 3 months before: 0
Major and severe incidents, Oct 2026
Average of the 3 months before: 0
All incidents
2 of 707 incidents, all dates
How this is counted
Counted by the date each incident was first reported. The bar that holds today is marked to date and is not compared. Every bar and number opens the incidents it counts.
Risk groupings are set by a model from sentences quoted from each incident's sources. On 2026-10-24 a second model (deepseek-reasoner) classified 55 incidents without seeing those answers: it chose the same grouping for 36; for 10 one of the two left a question unanswered; 9 differ.
| Date reported | Incident | Risk | AI developer | Severity |
|---|---|---|---|---|
| 2026-06-01 happened 2026-05-10 |
Marimo notebook flaw (CVE-2026-39987) exploited in LLM-agent-driven intrusion Preliminary | Data leaked or sent out Security compromise | Not stated | major |
| 2026-05-29 happened 2026-05-10 |
Attackers use LLM agent for post-exploitation after Marimo exploit Preliminary | AI used to run a cyber attack Security compromise | Not stated | major |
Browse incidents by
Risk grouping: AI used to exploit · Misuse by its own users · AI attacked · Not yet classified · AI acting on its own · AI failing in normal use · Harm by design · Supply chain attack
Kind of failure: Security compromise · Harmful output · Operational error · Other kind · Hallucination · Agent misalignment · Privacy leak · Physical safety · Bias or disparate impact · IP infringement
Developer: OpenAI · Anthropic · xAI · Tesla · Google · Meta · Microsoft · Waymo · Perplexity · DeepSeek · Facewatch · Alibaba · Amazon · Baidu · Character Technologies · Character.AI · GitHub · Peter Steinberger · Zoox · Amazon (AWS) · Amazon Web Services · Anthropic PBC · Atlassian · Aurelian · Avride · AWS · Bondu · Butterfly Effect · Character Technologies, Inc. · Cline · Codeway · Cognitec · Context.ai · Cursor · Deloitte · DreamX · Energize Lab · Figure AI · Flock Safety · Hugging Face · Instinct · Liquid AI · Lovable · McKinsey & Company · Moltbook · Moonshot AI · Nous Research · Open-source projects behind Strix, Cairn and Hermes · OpenClaw project (maintainer Peter Steinberger) · Perplexity AI Inc. · Replit · Salesforce · Shield AI · SpaceXAI · Tencent · Trezor · TypeSafe AI · Unitree Robotics · Univerads Technology Limited · Virtix Health · Wiz · X.AI LLC · XBOW · Xiaomi · Z.AI · Z.ai (Zhipu AI) · Zoox Inc.
Model: Claude · Grok · ChatGPT · Claude Code · Gemini · GPT-4o · Mythos 5 · Codex · DeepSeek · GPT-5.6 Sol · Muse · Claude Sonnet · Microsoft 365 Copilot · Astra · ChatGPT Atlas · Claude Mythos 5 · Claude Opus · Claude Opus 4.6 · Claude Opus 5 · Full Self-Driving · Gemini 3.1 Pro · GLM-5.3 · Manus · Model Y · OpenClaw · Tesla Model 3
Agent or product: Claude Code · OpenClaw · ARTEX · ChatGPT · Hermes Agent · Codex · CyberStrikeAI · AI agent · Amazon Bedrock AgentCore · Character.AI · chatbot · Cursor · Gemini CLI · humanoid robot · OpenAI agents · OpenAI Codex CLI
Tools it had: shell · browser · email · code_execution · file_read · file_write · network · bash · calendar · internet · python · email_read · exploitation · internet_access · mythic · network_access · package_install · api · api_calls · aws_api · code_generation · contacts · credential_access · file_edit · file_upload · files · Google Workspace APIs · messaging · messaging platforms · network scanning · powershell · reconnaissance · search · shell access · ssh · terminal · web search · web_browsing · web_scraping
How much it acted alone: Acted alone, unwatched · Acted alone, watched by a person · A person approved each action · It suggested; a person acted
Sector: Information · Finance and Insurance · Public administration · Educational Services · automotive · education · Retail trade · social media · Arts, Entertainment, and Recreation · government · Health care and social assistance · Legal services · Professional, Scientific, and Technical Services · software development · Software Publishers · Law Enforcement · Motor Vehicle Manufacturing · technology · artificial intelligence · banking · Cloud computing · Healthcare · higher education · legal · news publishing · Retail · software-as-a-service · transportation · Transportation and Warehousing · Aerospace and defense technology · AI app-building software development platform · AI developer tools · AI model development / evaluation · AI platform / intermediary services · AI research and development · AI safety evaluation (government research) · AI safety research · anime streaming · Artificial intelligence research and development · Autonomous ride-hailing (robotaxi) · Autonomous vehicle / robotaxi transportation · autonomous vehicles · blockchain analysis / financial technology · cloud computing / platform services · Cloud computing services · consulting · Consumer electronics / software · Consumer hardware / smartphone manufacturing · critical infrastructure and e-commerce · CRM software · cryptocurrency · Cybersecurity · decentralized finance (lending protocol) · Defense contracting · direct-to-consumer toilet paper deliveries · Emergency call centers (911 dispatch) · financial services · Fitness and recreational sports centers · Food Manufacturing · Government AI safety and security evaluation · government AI safety evaluation · Government and electric power · government contracting and immigration enforcement · grocery retail · hardware wallet manufacturing · Health insurance / public healthcare administration · Healthcare and government · hobbyist networking · Industrial automation, electronics manufacturing, semiconductors, pharmaceuticals, shipping, tire manufacturing, petrochemicals, and data center services · Information (software publishing / AI services) · Information Security research · Insurance · IT services (managed service providers) · Manufacturing (robotics) · matrimonial services · Messaging and social media platform · military · news media · online gambling · Open knowledge platform / nonprofit encyclopedia · open source software · photography · Police / public safety · Police and law enforcement · Policing · political campaigns · Political campaigns and elections · Public administration and courts · Public administration and defence · Public administration and government · railway · Real estate · Real estate (hobbyist manufacturing) · Recreation and outdoor travel (consumer chatbot use) · religious organisation · Research and development in AI safety · Retail Trade; Transportation and Warehousing; Finance and Insurance · ride-hailing · Ride-hailing / autonomous vehicle transportation · Ride-hailing and autonomous vehicle passenger transport · robotics · Scientific research and light source user facility · Social media / AI platform · Social media / online content publishing · Social media / technology platform · social media platform · Social media platform (content moderation) · Software / AI coding tools · Software development (AI vibe-coding platform) · Software development / AI · Software development / AI tooling · Software development / coding tools · Software development / cybersecurity research · Software development / information sector · Software development tools · Software publishing / npm ecosystem · software studio · Taxi and Limousine Service · theme parks · Ticketing and live entertainment · Toy and consumer electronics manufacturing
Severity: moderate · major · minor · severe · negligible
Where it happened: In production · In evaluation · In research · In training