AINewsnow

Study: poisoning 0.1% of pretraining data implants a working RCE backdoor in an AI agent

📅 2 November 2026 • arXiv / Hacker News A study posted to arXiv on October 27, 2026 describes a result that changes the threat model for AI agents: poisoning a small share of pretraining data can implant a working backdoor in an agent. The paper, numbered 2610.37741 , is authored by Xinyan Hu and…

Read the full story at DEV Community — Machine Learning ↗

Timeline · 1 report

  1. 2026-11-02 23:16 · DEV Community — Machine Learning
    Study: poisoning 0.1% of pretraining data implants a working RCE backdoor in an AI agent

More stories

  1. OpenAI appoints Dan Driscoll as Senior Counselor for National Security Modernization — OpenAI News
  2. Marvell to acquire Lumentum in $115 billion all-stock AI data-centre deal — DigiTimes Asia
  3. Anthropic reports $17B quarterly revenue, $900B in compute commitments ahead of IPO — Reuters AI
  4. OpenAI launches Grid Protectors Program to cut data center power use during grid strain — OpenAI News
  5. Anthropic confirms new Claude Code limits to prevent recreating existing video games — Engadget
  6. Israeli startup 42AI exits stealth with $62.5M seed for behavioral world model — TechCrunch AI
  7. OpenAI to merge ChatGPT's Chat and Work tabs into a single desktop experience — TechCrunch AI
  8. October slide in AI and tech stocks tests broader US market's resilience — Reuters AI

Get the daily brief of stories like this at 6:30 every morning →