# Three new papers move prompt-injection defense for agents to the point of action

APEX enforces checks where agents call tools, MCP servers and skills, DIBench tests mobile GUI agents against deceptive screens, and Proxy Confidence uses a surrogate model's log-probabilities to audit closed agents.

- Published: 2026-10-07T06:04:58.881Z
- Canonical: https://polylog.news/ai/2026-10-07/three-new-papers-move-prompt-injection-defense-for-agents-to
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [arXiv cs.CR (APEX)](https://arxiv.org/abs/2610.06966), [arXiv cs.CR (DIBench)](https://arxiv.org/abs/2610.06898), [arXiv cs.AI (Proxy Confidence)](https://arxiv.org/abs/2610.03894)

Three preprints posted to arXiv on October 7 address the same problem using different methods. Agents now read untrusted content and act on it, and current defenses do not inspect the action itself. APEX (Active Protection at Execution Boun…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-10-07/three-new-papers-move-prompt-injection-defense-for-agents-to (subscription information: https://polylog.news/pricing).