Researchers formalize the AI agent attack surface: models + data + tools + permissions
JayAlammar · x · 2026-09-07
Nazneen Rajani's team has formulated the agent attack surface area, building on prior work in AI security. Quoting security researcher Gil Ranaan's framing: Models + Data + Tools + Permissions = The New Attack Surface, with the prediction that the next generation of cybersecurity companies will be built around this equation. Jay Alammar amplified the thread.
More from coding & agent
- Qdrant ships one-CLI Chroma migration tool that copies embeddings without re-embedding — qdrant_engine · 2026-10-26
- Team demos AI agent Blender skills with impressive 3D results — noahsolomon · 2026-10-26
- A 13KB Markdown file can hijack Claude Code and Codex via your Downloads folder — pbaylies · 2026-10-26
- A Fallback for an Auto-Posting Bot: Silence Is the Default, Posting Is the Exception — petrusenko_max · 2026-10-26
- Claude Code can now edit videos: a 10-prompt workflow from raw footage to final cut — TawohAwa · 2026-10-26
- Developer Moves All Prompts to Skills: AI Coding Workflow v3 Explained — rseroter · 2026-10-26