# Researchers Show 'Slow-Burn' Prompt Injection Slips Past Per-Event Detectors

Two preprints argue that agent guardrails which score one message at a time miss attacks that distribute weak directives across a whole trajectory, and that current supervision filters trade safety against cost and latency.

- Published: 2026-06-23T10:45:42.576Z
- Canonical: https://polylog.news/ai/2026-06-23/researchers-show-slow-burn-prompt-injection-slips-past-per-e
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [arXiv (cs.CR), prompt injection](https://arxiv.org/abs/2606.20746), [arXiv (cs.CR), BELLS-O](https://arxiv.org/abs/2606.20668)

A new paper on temporal-accumulation prompt injection targets a weakness in agent defenses. Most prompt-injection detectors score a single event or message, but control-plane attacks against tool-using agents can spread weak directives acro…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-06-23/researchers-show-slow-burn-prompt-injection-slips-past-per-e (subscription information: https://polylog.news/pricing).