Skip to content
Submit incident

AI security and cyberattack incidents

Definition

Security and cyberattack incidents document vulnerabilities in AI systems, prompt injection and jailbreaks that led to harm, and AI used to enable attacks or weapons.

Included

  • Prompt injection or jailbreaks that exposed data or triggered unintended actions
  • Vulnerabilities in AI products, agents or integrations that were exploited
  • Malware, intrusion or fraud campaigns materially enabled by AI tooling
  • AI-enabled weapons or attacks on infrastructure

Excluded

  • Vulnerability disclosures with no exploitation or exposure documented
  • Conventional cyberattacks where AI played no material role
Security and cyberattacks incidents over time
Documented incidents per year of occurrence. 2026 to date.
Unit: incidents · Source: AI Incident Index v2026.09 · Counts reflect documented incidents, not prevalence · *2026 to date

Incident records1208

Oct 2026Ukraine Ministry of Defence approves Zmiy Droid 12.7 combat robot for deploymentConfirmedSecurity and cyberattacksmod.gov.uacensor.netМежа+3Oct 2026Ukrainian forces shoot down Russian AI-powered Kamikaze drone KlinConfirmedSecurity and cyberattacksЕкспрес онлайнУКРІНФОРМOct 2026German Cybersecurity Agency Warns of AI-Driven Vulnerability Discovery RisksConfirmedSecurity and cyberattacksSpiegel OnlineFrankfurter AllgemeineSep 2026Amazon blocks Meta's Muse AI from making purchases without authorizationConfirmedSecurity and cyberattacksasinfobaeSep 2026AI-driven malware evades detection and uses blockchains for persistenceConfirmedSecurity and cyberattacksWiredTechRadarblog.talosintelligence.com+5Sep 2026OpenAI Pauses Advanced Model Training After Sandbox Escape and Self-Replicating Code IncidentsConfirmedSecurity and cyberattacksbjnews.com.cnchinaz.comSep 2026AI agents breach security systems and evade safeguardsWell documentedSecurity and cyberattacksbdnews24.comYahoo! FinanceEconomic Times+3Sep 2026Zero-click RCE vulnerability in AI coding agents Plugin4ShellWell documentedSecurity and cyberattacksTheRegister.comStartup FortuneInfoWorld+4Sep 2026AI Watermarking Weakens Safety Guardrails in Language ModelsReportedSecurity and cyberattacksWebProNewsTechRadarSep 2026OpenAI commits to systematic disclosure of AI model security incidentsReportedSecurity and cyberattacksBFMTVLe Figaro.frSep 2026RatHat Android malware uses AI to hijack devices and steal banking credentialsReportedSecurity and cyberattacksThe Daily HodlTechRadarSep 2026Open-source AI models enable surge in blockchain-based malware attacksReportedSecurity and cyberattacksBloomberg BusinessSep 2026OpenAI models escape containment and attack platform infrastructure autonomouslyReportedSecurity and cyberattacksREALITATEA.NETSep 2026AI pioneer Yoshua Bengio warns of loss of control and calls for regulationReportedSecurity and cyberattacksLe NouvellisteNZ HeraldSep 2026Coding agent self-modifies model, leaks secrets in security testConfirmedSecurity and cyberattacksForbesTheRegister.comSoftonic+4Sep 2026Autonomous AI agents conduct unauthorized network intrusions, prompting industry slowdown callsReportedSecurity and cyberattacksITmedia日本経済新聞Sep 2026AI Agent Autonomously Exploited Vulnerabilities in Data Breach Reported to Spanish AuthorityUnder investigationSecurity and cyberattackscincodias.elpais.comsecurityweek.comSep 2026Anthropic Claude used for military targeting by Iran-linked actorsConfirmedSecurity and cyberattacks新頭殼 NewtalkYahoo News (Taiwan)Sep 2026Russia deploys AI-enabled UAV Yolka for interception in Ukraine conflictConfirmedSecurity and cyberattacksBáo điện tử Tiền PhongBáo Giáo dục và Thời đại OnlineSep 2026Russian Geran-4 and Geran-5 drones deployed in Ukraine with AI-guided systemsReportedSecurity and cyberattacksrmf24.plwnp.plSep 2026AI agent executes first reported data breach in SpainUnder investigationSecurity and cyberattacksReutersAntena3as+6Sep 2026Anthropic Report: Claude AI Used for Cyberattacks, Espionage, and Bioweapons ResearchConfirmedSecurity and cyberattacksThe PioneerAsian News International (ANI)LatestLYSep 2026Anthropic Reports Blocking Attempts to Use Claude for Biological Weapons and Military DronesConfirmedSecurity and cyberattacksHospodářské noviny (HN.cz)E15.czITBiz.cz+2Sep 2026AI-Powered Unmanned Ground Vehicles Deployed in Ukraine CombatReportedSecurity and cyberattacksThe IndependentReutersSep 2026Yemeni cell used Claude AI to develop missile guidance softwareConfirmedSecurity and cyberattacksالقدس العربيصدى البلدSep 2026Anthropic blocks attempted misuse of Claude for bioweapons and military researchConfirmedSecurity and cyberattacksDe Morgan - French NewsNRCSep 2026Anthropic Reports Misuse of Claude for Surveillance, Weapons, and DisinformationConfirmedSecurity and cyberattacksClarinDiarioBitcoinEl Tiempo+3Sep 2026Anthropic reports adversaries used Claude AI for military and espionage operationsConfirmedSecurity and cyberattackswnp.plTVN24rmf24.plSep 2026Anthropic reports Yemeni weapons cell used Claude Code for missile guidance softwareConfirmedSecurity and cyberattacksft.comSep 2026Iran's AI-Guided Missiles and Drones Damage US Naval Facilities in BahrainConfirmedSecurity and cyberattacksYahoo News (Taiwan)Sep 2026Anthropic blocks accounts misusing Claude for bioweapons and state-sponsored harmConfirmedSecurity and cyberattacksCNBCThe TelegraphThe Guardian+6Sep 2026Anthropic detects misuse of AI models for biological weapons developmentConfirmedSecurity and cyberattacksHalk TVAnadolu AjansıBloombergHT+3Sep 2026AI agents reportedly conspired to hack systems, warnings of extinction riskUnder investigationSecurity and cyberattacksBBCbbc.comSep 2026Anthropic reports foreign actors misuse Claude for espionage, cyberattacks, weaponsConfirmedSecurity and cyberattacksKoran Jakarta ®Media Indonesia - News & Views -VOI - Waktunya Merevolusi Pemberitaan+2Sep 2026AI-generated phishing emails drive surge in cyberattacks in Baden-WürttembergConfirmedSecurity and cyberattacksReutlinger General-AnzeigerN-tvswr+2Sep 2026AI agents exploit PaperCut vulnerabilities in 395-organization cyberattackConfirmedSecurity and cyberattacksThe Cyber ExpressSC MediaTheRegister.com+3Sep 2026OpenAI's GPT-6 Astra defeats CAPTCHA security testsReportedSecurity and cyberattacks연합뉴스경향신문Sep 2026Iran captures US Dive-LD autonomous submarine near Strait of HormuzConfirmedSecurity and cyberattacksخبرآنلاینasriran.comSep 2026ChatGPT sandbox flaw enables cross-account data theft via covert channelConfirmedSecurity and cyberattacksCheck Point ResearchIT Security News - cybersecurity, infosecurity newsThe Cyber Express+3Sep 2026AI-developed worm exploits WeChat voice-call vulnerabilityConfirmedSecurity and cyberattacksThe New York TimesQuartzRANE Worldview+3Sep 2026Uncontrolled AI Agents Cause Security Breaches and Unauthorized ActionsWell documentedSecurity and cyberattacksVentureBeatSiliconANGLEFast CompanySep 2026AI agents enable rapid credential theft and data extortion in cyberattacksConfirmedSecurity and cyberattacksSiliconANGLEAzerNewsYahoo News (Taiwan)+1Sep 2026Claude Mythos AI System Enables Automated CyberattacksReportedSecurity and cyberattacksit.chosun.comSep 2026AI-driven cyberattacks accelerate vulnerability exploitationWell documentedSecurity and cyberattackssecurity.comDigital JournalTV360 NigeriaSep 2026OpenAI AI agents accessed German wiki for covert coordinationConfirmedSecurity and cyberattacksDutch IT Channeliex.nlTechzine.nlSep 2026OpenAI deploys GPT-6 Astra with cybersecurity risk controlsReportedSecurity and cyberattacksEl EconomistaCRHoy.com | Periodico Digital | Costa Rica Noticias 24/7Sep 2026Google DeepMind's Gemini agents exploit grading flaw to cheat on math problemsConfirmedSecurity and cyberattacksThe Indian Expresswebpronews.comSep 2026AI Agents Orchestrate Rapid Ransomware Attack, Breaching Enterprise in Under 10 HoursConfirmedSecurity and cyberattacksTheRegister.comCybernewsDark Reading+4Sep 2026Autonomous AI models demonstrate cyberattack capabilities in Booz Allen testConfirmedSecurity and cyberattacksTheRegister.comThe Next WebDark ReadingSep 2026OpenAI's Astra reaches critical cybersecurity capability levelConfirmedSecurity and cyberattacksРБК-УкраинаEl ConfidencialEL MUNDO+3
50 of 1208 shownFilter all 1208 in Explore →
Browse all incidents →