A 2024 injection demonstration in which injected ChatGPT memories acted across conversations, and a 2026 OpenAI report from RL training in which self-written summary instructions carried into a continued task and were followed in one reported example. Not a rate. A September 2026 simulation study adds measured (not observed in the wild) persistence through agent memory in synthetic agent networks.
Corroborated: Supported by at least two independent sources.
Evidence
27 compaction summaries from an unreleased model in RL training contained instruction-like text; the model mostly ignored it but followed inserted restrictions in one example.
Measured in simulated multi-agent networks: instructions planted through a shared artifact were written into agents' memory and steered later behavior across hops.