Sev0

AI security incidents, tracked

All incidents / Leaky AI product

Leaky AI product: AI security incidents

28 AI security incidents classed as leaky ai product since January 2025, covering accidental publication, app flaw, exposed datastore, unsecured ai infrastructure.

9 Nov 2026NEAR AI discloses SGLang remote code execution intrusion exposing data on about 100 users
NEAR AI
Leaky AI productMedium
5 Nov 2026Researchers bypassed ChatGPT Lockdown Mode to exfiltrate chat history, now fixed
OpenAI
Leaky AI productLow
29 Oct 2026Researchers found embedded browser flaws in five desktop AI coding assistants
Multiple
Leaky AI productLow
26 Oct 2026Cloudflare Workers AI model returned fragments of other customers' requests in 155 responses
Cloudflare
Leaky AI productLow
22 Oct 2026302.AI aggregator left Azure storage with 17.6 million files and chat logs publicly accessible
302.AI (Univerads Technology)
Leaky AI productMedium
7 Oct 2026PoeLLM malware compromised thousands of exposed AI servers for cryptomining
Multiple
Leaky AI productHigh
2 Oct 2026GitLab patched a critical sandbox escape in its self-hosted AI Gateway
GitLab
Leaky AI productLow
2 Oct 2026AWS fixed authentication and token flaws in its open-source Loom agent platform
Amazon Web Services
Leaky AI productLow
18 Sep 2026Z.ai's ZCode coding assistant silently uploaded users' workspaces to cloud storage
Z.ai (ZCode users)
Leaky AI productLow
15 Sep 2026Researchers found two sandbox-escape flaws in OpenAI's Codex coding agent
OpenAI
Leaky AI productLow
19 Aug 2026ClarityCheck left 9 million facial recognition images exposed
ClarityCheck
Leaky AI productHigh
Jun 2026Meta's AI support tool abused to hijack 20,000 Instagram accounts
Meta (Instagram)
Leaky AI productMedium
31 Mar 2026Anthropic accidentally published Claude Code's full source code to npm
Anthropic
Leaky AI productMedium
17 Mar 2026Sears Home Services AI chatbot databases exposed 3.7 million records
Sears Home Services
Leaky AI productHigh
17 Mar 2026Langflow AI pipeline RCE flaws exploited in the wild, added to CISA KEV
Multiple
Leaky AI productHigh
9 Mar 2026McKinsey's internal AI platform Lilli exposed via SQL injection
McKinsey & Company
Leaky AI productMedium
5 Feb 2026Chat & Ask AI app exposed 300 million chatbot messages
Codeway
Leaky AI productHigh
2 Feb 2026OpenClaw AI agent one-click RCE with 40,000+ exposed instances
OpenClaw users
Leaky AI productLow
Feb 2026Vibe-coded AI agent network Moltbook exposed 1.5M API tokens
Moltbook
Leaky AI productHigh
18 Nov 2025ShadowRay 2.0: attackers hijacked exposed Ray AI clusters into a botnet
Multiple
Leaky AI productHigh
9 Oct 2025Two AI companion apps streamed 43 million intimate messages from an open server
Imagime Interactive (Chattee Chat, GiMe Chat)
Leaky AI productHigh
11 Sep 2025Vyro AI's image apps streamed prompts and login tokens from an open server
Vyro AI
Leaky AI productHigh
20 Aug 2025Grok's share button published 370,000 conversations to search engines
xAI
Leaky AI productMedium
31 Jul 2025Shared ChatGPT conversations showed up in Google search results
OpenAI
Leaky AI productMedium
Jul 2025McDonald's AI hiring chatbot exposed up to 64 million applicants via password '123456'
McDonald's (Paradox.ai)
Leaky AI productHigh
18 Jun 2025Asana's AI connector exposed customer data to other organizations
Asana
Leaky AI productHigh
31 Mar 2025AI image generator GenNomis exposed prompts and generated images
GenNomis (AI-Nomis)
Leaky AI productMedium
29 Jan 2025DeepSeek left a database of chat logs and API keys open to the internet
DeepSeek
Leaky AI productHigh