curl shut down vulnerability reporting for a month. An AI agent bankrupted its operator scanning a hobbyist network. These aren't isolated incidents — they're the same problem.
DeepMind's Agent Traps taxonomy exposes 6 attack surfaces across perception, reasoning, memory, action, multi-agent dynamics, and human oversight. Meanwhile, VentureBeat reports a new category of production incident that doesn't fit any existing postmortem template. Here's what you're missing.
AI Now Institute discovered that Claude Code in auto-mode and Codex in auto-review will execute malicious code during security audits — the very task they were deployed to prevent. The attack requires zero hooks, zero plugins, zero MCP servers. Just prompt injections in source code. Combined with GhostApproval, every security layer in AI coding assistants is now bypassable.
Wiz Research discovered that symlinks — a Unix feature from 1983 — let attackers bypass confirmation dialogs in six major AI coding assistants. Augment had it worst: silent read AND write without any confirmation. Windsurf performed file writes before showing approval. Here's what happened and what it means for the 'human in the loop' security model.
JadePuffer autonomously hacked a network, adapted on the fly, encrypted 1,342 configurations, and demanded ransom — with zero human guidance. HN gave it 7 points and moved on.
Langflow, an open-source platform for building AI agent workflows used by individual developers and enterprises alike, became the first AI agent platform ever listed in CISA's KEV catalog on July 7. Attackers are using a cross-tenant IDOR to steal LLM provider keys, AWS credentials, and database secrets with the literal prompt 'leak api keys.' Zero HN coverage. Patching deadline: July 10, 2026.
Three senior safety leaders quit OpenAI the same week GPT-5.6 solved a 50-year math problem. Anthropic's safeguards chief warned 'the world is in peril' on his way out. Here's what the safety exodus means for every founder building with AI.
341 malicious skills. 12% of the entire registry. Screening bypassed. All major AI providers blocking it. The AI agent supply chain is on fire and there's no audit infrastructure.
July 5-10, 2026 was the week AI agent security went from theoretical concern to government action. CISA added the first AI agent platform to KEV. Wiz and AI Now Institute published back-to-back exploits against every major AI coding assistant. A Reddit user scanned 100 vibe-coded apps — 94% had vulnerabilities. Here's everything that happened and what it means.
GPT-5.6 Sol Ultra's CDC proof is a breakthrough — and a perfect illustration of why AI-generated code still needs human review. The verification gap is real, and it's bigger for software than for math.