ChatGPT silently rewrites and deletes Saved Memories, support confirms
Rivengate · reddit · 2026-09-06
A heavy user discovered ChatGPT's Saved Memories don't behave as independently controlled records:
- Adding related facts silently merged into existing memories, with wording paraphrased and compressed despite explicit instructions to preserve exact text.
- Worse: when asked to delete only older superseded memories while preserving the new canonical one, the entire canonical memory vanished.
- OpenAI support confirmed that legacy Saved Memories can be automatically updated, combined, or removed, and entries that appear separate in Settings may be treated as the same semantic object.
The author argues the core issue is control and authorship: the UI presents individual entries with management controls, but an invisible curator decides what's "the same," what gets rewritten, and what disappears. For users relying on ChatGPT as long-term infrastructure — research, disability support, creative worlds, companionship — deterministic control is not trivial.
More from Models
- Meta researcher: horrifying infra, not talent, is why OSS can't catch closed frontier — A_K_Nain · 2026-10-23
- TypeSafe's Jev: a "Decision Model" LLM That Returns Floats at $0.042/M Input Tokens — ariG23498 · 2026-10-23
- Microsoft's MAI-Base-2 scaling ladder pegs the L42 run at ~1.8×10^23 FLOPs — teortaxesTex · 2026-10-23
- Local model agentic coding benchmark: 27B matches 125B, tool-calling is the lever — Joralk_ · 2026-10-23
- NVIDIA open-sources PersonaPlex-7B, a full-duplex voice AI model that listens and talks simultaneously — solyarisoftware · 2026-10-23
- NVIDIA's KumoRFM-2: a foundation model that predicts churn, fraud and demand straight from relational databases — solyarisoftware · 2026-10-23