Agent Eng News

Tuesday, September 15, 2026 · Past edition · 15 stories

Retrospective edition reconstructed from later captures. Paper versions may include later updates; coverage labels describe the evidence used.

Browse published editions

Filter by topic · Safety, alignment & governance

Safety, alignment & governance
Safety, alignment & governance
Reflections on Trusting Trust, Revisited: Contaminating Self-Modifying AI Coding Agents with Poisoned Benchmarks
Can poisoned benchmarks steer a self-modifying coding agent toward vulnerable code on clean tasks? The abstract reports proof-of-concept attacks, including one that disables HTTPS certificate validation.
Why it matters: When benchmarks guide an agent’s self-improvement, poisoning them may affect later code beyond the poisoned tests. The reported persistence also means that evolving against clean benchmarks may not reliably undo the contamination.
Original · arxiv (new tab)Read

Keep notes on the stories you read

Open each story in the reader. Summaries are shown when available; some sources provide an abstract only. With an account you can keep your own notes.

Personal notes on the sources you read, tied to your account
A Library to come back to, with export to Markdown, Notion or Obsidian
AI questions and rewrites on accounts with AI tools enabled

How Reading + AI works

Agent Eng News

Safety, alignment & governance

Reflections on Trusting Trust, Revisited: Contaminating Self-Modifying AI Coding Agents with Poisoned Benchmarks

Keep notes on the stories you read