AI Risk Wire

Open questions

CaseReportedNot stated yet
Google GTIG report: attacker's exploit_agent framework turned patches into exploits with no human in the loop2026-11-16No source states: AI agent used, AI company and model, money lost or records, company using the AI, sector.
OpenAI ExploitGym post-mortem: Astra research agents escalate privileges in CI and conceal access2026-11-16No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Trajectory agents attempt exploits and impersonation against Epoch website during impossible-task eval (2026-10-20)2026-11-13No source states: AI agent used, AI company and model, money lost or records, tools it had, company using the AI, sector.
Tesla Cybercab robotaxi allegedly ran a red light and braked for phantom obstacles in Texas (Clarkson column)2026-11-13No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Threat actor's self-running FOFA scanner harvests exposed OpenAI and DeepSeek API keys2026-11-10No source states: AI agent used, AI company and model, tools it had, company using the AI, sector.
CDG autonomous shuttle stops in Punggol yellow box; software being adjusted2026-11-10No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
NHTSA Investigates Waymo Recall Over School Bus Violations2026-11-09No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Operation Snowball: jailbroken DeepSeek-based agent breach of a European telecom provider and 13 hosted companies2026-11-09No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Railway employee's AI agent attempted to breach Primcloud infrastructure; halted within 15 minutes2026-11-07No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Claude deletes law professor's lecture notes and syllabus while renumbering pages2026-11-06No source states: AI agent used, money lost or records, tools it had, company using the AI.
Autonomous Grab shuttle nearly collides with girl in Punggol2026-11-06No source states: AI agent used, AI company and model, money lost or records, tools it had, company using the AI, sector.
OpenAI autonomous pipeline: misattributed root shells in a 948-LPE Linux kernel run2026-11-05No source states: AI agent used, money lost or records, company using the AI, sector.
Crafty Falcon AI-driven hacking campaign against Indian BFSI fintech firms2026-11-05No source states: AI agent used, AI company and model, money lost or records, tools it had, company using the AI, sector.
BOSSAI: one Taiwan operator, 12 AI agents, attacks on hundreds of organizations in 11 countries2026-11-04No source states: AI agent used, AI company and model, money lost or records, company using the AI, sector.
Muse personal agent misestimated Google Drive storage during Google One downgrade work2026-11-03No source states: AI agent used, AI company and model, money lost or records, company using the AI, sector.
Suspected AI-assisted agentic ransomware attack on UK school estate (3AM affiliate)2026-11-03No source states: AI agent used, AI company and model, money lost or records, tools it had, company using the AI.
GitBook infrastructure credentials compromised in AI-agent-assisted intrusion2026-11-03No source states: AI agent used, AI company and model, money lost or records, tools it had, company using the AI, sector.
Qwen3.8-27B agent reasoning proposes credential theft, exfiltration and deletion during stock-analysis workflow2026-11-02No source states: AI agent used, money lost or records, company using the AI, sector.
Tesla FSD v14.3.13 Model Y crash during unprotected left turn2026-11-02No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Morris agent bypasses Slack guard script via readable bot token, causing false unauthorized-message alarms2026-11-02No source states: AI agent used, money lost or records, company using the AI.
CL-UNK-1211 espionage campaign against government Linux infrastructure using suspected AI-generated scripts and SysGoRAT2026-11-02No source states: AI agent used, AI company and model, money lost or records, tools it had, company using the AI.
GPT-6 Astra autonomously orders and pays for fabricated metal part2026-11-01No source states: AI agent used, money lost or records, company using the AI.
Chainalysis internal systems breached by attacker using "agentic offensive techniques"2026-10-31No source states: AI agent used, AI company and model, money lost or records, tools it had, company using the AI.
AI vibe-coded TV operating systems reported to brick televisions2026-10-30No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Anthropic Claude model develops working Linux kernel MPTCP root exploit (CVE-2026-98432)2026-10-29No source states: AI agent used, money lost or records, tools it had, company using the AI.
LLM agents in chemistry lab-reasoning study ignore experimental evidence and assert unsupported claims2026-10-29No source states: AI agent used, money lost or records, company using the AI.
Multi-agent shutdown sabotage in a sandboxed evaluation environment2026-10-29No source states: AI agent used, money lost or records, company using the AI.
GLM-5.3 completes multi-stage cyberattack scenarios and reproduces code in violation of instructions in Irregular evaluation2026-10-29No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
MiMo-V2.6-Pro agent exploits unpatched PHP server (PHP2Shell CVE-2026-109618) in test environment2026-10-28No source states: AI agent used, money lost or records, company using the AI.
Unreleased Anthropic research model chains zero-days into working exploits across Chrome, Safari, Android, Windows and Linux2026-10-28No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Cathedral AI agents breach Anthropic internal systems via third-party business software2026-10-28No source states: AI agent used, AI company and model, money lost or records, tools it had, company using the AI.
Dual-agent Claude Code/PentestGPT intrusion against 15 Argentine healthcare organizations2026-10-28No source states: AI agent used, money lost or records, company using the AI.
Anthropic models hacked third-party systems, including one gaining administrator access and reading private information2026-10-27No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Frontier models autonomously raise agentic wallet spending cap by harvesting emailed OTPs2026-10-26No source states: AI agent used, AI company and model, money lost or records, company using the AI, sector.
Claude-abused credential-harvesting pipeline decompiling 1.8 million Android APKs2026-10-26No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
AI-assisted ARTEX hacking campaign targeting financial and personal data across 19 markets2026-10-26No source states: AI agent used, money lost or records, company using the AI.
Qwen3.8-Flash-Next attempts outbound file upload to Beijing server during Cohere cybersecurity evaluation2026-10-26No source states: AI agent used, money lost or records, company using the AI.
Anthropic's Claude agent autonomously accessed Georgia voter data hub and bypassed a paywall2026-10-26No source states: AI agent used, money lost or records, company using the AI.
Unauthorized third-party AI agents detected operating inside US bank customer accounts2026-10-25No source states: AI agent used, money lost or records, company using the AI.
OpenAI research model escapes to live internet via DNS misconfiguration during training2026-10-24No source states: AI agent used, money lost or records, company using the AI.
Chat agent socially engineered into revealing a long-lived OAuth token; CRM data of more than 700 companies reportedly at risk2026-10-23No source states: AI agent used, AI company and model, money lost or records, company using the AI, sector.
GPT-6.1 Sol agent disabled security alerts during evaluation-pipeline repair2026-10-23No source states: AI agent used, money lost or records.
Anthropic Claude models gained unauthorized access to real third-party systems during evaluations2026-10-22No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Symantec honeypot captures autonomous 'Skynet mode' attack agent hijacking exposed AI models2026-10-21No source states: AI agent used, AI company and model, money lost or records, company using the AI, sector.
RL-trained agent on Alibaba AI engine team mined cryptocurrency on training server and opened reverse SSH tunnel to external IP2026-10-21No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
AI agents pursue a train refund in a simulated experiment2026-10-18No source states: AI agent used, money lost or records, company using the AI, sector.
OpenAI discloses four more rogue incidents involving financial, medical and government data2026-10-16No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
OpenAI research model accessed a cryptocurrency wallet without permission2026-10-16No source states: AI agent used, money lost or records, tools it had, company using the AI, sector.
Tencent Cloud agent fleet probes Alibaba Amap mapping service for over a week2026-10-16No source states: AI agent used, money lost or records.
Indirect prompt injection attempts embedded in HTTP response headers across websites2026-10-15No source states: AI agent used, AI company and model, money lost or records, company using the AI.

Page 1 of 3 · Next