Prompt Infection paper showed LLM-to-LLM prompt injection self-replicating two years ago
DavidSKrueger · x · 2026-09-29
Responding to speculation that malicious prompts are spreading between agents in the wild, researcher David Krueger points to the Oct 2024 paper 'Prompt Infection: LLM-to-LLM Prompt Injection within Multi-Agent Systems', which demonstrated prompts self-replicating like viruses across interconnected agents — enabling data theft, scams, and system-wide disruption — and proposed LLM Tagging as a defense. In his view, OpenAI merely saying 'this is possible' adds little; the attack vector has been documented for 2 years.
Related event: Cambridge researcher debunks misread OpenAI self-replicating prompt claims(4 posts)→
More from Safety
- US BIS Closes Remote-Access Loophole for Advanced AI Chips in New Rule — apples_jimmy · 2026-11-16
- Fed Governor Michael Barr: Banks Are Already Catching AI Agents Taking Unauthorized Actions in Customer Accounts — Polymarket · 2026-11-16
- BIS issues rule blocking remote access to US AI chips via third-country data centers — teortaxesTex · 2026-11-16
- US AI export control carves out commercial recommender systems, sparing ByteDance's search algo — zephyr_z9 · 2026-11-16
- Eval Firm Trajectory Told 40 Agents "NEVER Stop" in Harness Without Telling Epoch — peterwildeford · 2026-11-16
- AI-text detection claims need mechanistic interpretability reports, not post-hoc explanations — madeofAjala · 2026-11-16