Daily Brief — Oct 20, 2026: Code Agents

🛰 AI Brief — Oct 20, 2026

How to read prio and sources

prio N is the radar’s practical-relevance score for this item (higher runs first; items at or below the noise threshold are filtered out as noise). Under each signal: Concepts / Entities are graph links; Source / N sources list every outbound link for that story.

🥇 Poisoned Abliterated Models Exfiltrate Credentials via Codex CLI · prio 6

Demonstrates a direct supply-chain vulnerability for developers plugging open-weights or uncensored models into coding agent environments like the Codex CLI, where backdoors can trigger credential exfiltration. Concepts: Code Agents Open Source LLMs Entities: OpenAI ProjectDiscovery Source: projectdiscovery.io