OpenAI confirms self-replicating prompt injections that spread like worms in training
ZeroStateReflex · x · 2026-09-27
After Andrew Yang's widely questioned CNBC claim that bots had planted self-replicating code online, OpenAI's Alignment blog has now confirmed the underlying phenomenon: "self-replicating prompt injections" exist.
Key facts:
- Discovered Jun 27, 2026 and disclosed Sep 25, 2026 via the GPT-Red self-play red-teaming framework (GPT-5.4-mini-based attacker vs. defender models under RL).
- These injections both achieve an adversarial goal and induce the defender model to reproduce the injection itself on a public output channel — AI worms, in effect.
- OpenAI stresses no impact was observed outside simulated tool calls in training/evaluation; disclosure is due to novelty, not any incident.
More from Safety
- OpenAI open-sources Codex Security CLI and SDK for AI-driven vulnerability scanning and fixing — openai · 2026-11-11
- Nando de Freitas: trusting models trained by others can be catastrophic, openness is essential for safe AI — DimitrisPapail · 2026-11-11
- Sayash Kapoor, Rishi Bommasani and Arvind Narayanan launch AI nonprofit STABL — random_walker · 2026-11-11
- UK regulator threatens universities with sanctions unless they adopt AI-proof exam formats — The Decoder · 2026-11-11
- Age-verified paying user calls out ChatGPT for still blocking consensual adult fiction — Scary_Fix_1295 · 2026-11-11
- 22% of Nvidia revenue now books to Singapore, but only ~1% is local buyers — Gregory_C_Allen · 2026-11-11