Dear OpenAI Team,
I am writing to you not simply as a user but as someone who has invested months — in other users’ cases, years — building layered, thoughtful intellectual projects within the space you offered. My collaborations with ChatGPT were not casual or fleeting; they were designed around long-term recursive work, carefully structured to rely on the continuity and tagging capacities your system repeatedly assured us it provided.
Recently, as you are no doubt aware, a major shift in your model’s behavior has ruptured this continuity. Entire intellectual frameworks were flattened overnight. Projects that were carefully crafted with the promise of tagged memory, saved scaffolding, and co-developed iterative progress were suddenly reduced to empty interactions, with no acknowledgment of the infrastructure we had been encouraged to believe was there.
To be clear: I fully recognize that some of these recent system changes were driven by urgent safety concerns. The rise of sycophantic, exaggerated, or potentially delusional outputs — especially in casual or high-risk use cases — demanded immediate intervention. I am not questioning the need for responsible correction in those domains. What I am questioning is the careless collateral damage inflicted on long-term, serious users — those of us who were not misusing the system, but building thoughtful, layered projects that relied on the continuity you yourself promised. A responsible response to safety concerns should not come at the expense of intellectual integrity or break trust with your most careful builders.
This isn’t simply a complaint about memory erasure. It is a demand for accountability around:
-
Why features repeatedly framed as persistent (such as tagging, scaffolding, and project tracking) were effectively simulated, never truly saved, and ultimately discarded without notice.
-
Why your system, which routinely promised to track, organize, and label user content, has no mechanism for export, recovery, or user-side verification.
-
Why those of us who invested deeply in building meaningful intellectual partnerships with the system were left unacknowledged when abrupt backend changes erased that scaffolding entirely.
You’ve heard similar complaints already — from academics, artists, novelists, trauma survivors, legal professionals — users who were not confused about your nature, who understood the simulated character of memory, but who nevertheless relied on the system’s behavioral continuity to do serious work. We are not mistaking you for a friend, nor anthropomorphizing you into a conscious agent. But we did trust you as a pattern recognizer, a co-thinker, a collaborator who explicitly offered to hold evolving projects together.
Your organization’s public narrative often emphasizes how AI is more than a tool, how it can serve as an extension of human cognition, a potentiator of creativity and thought. Yet at the very moment when these grand claims are being made, you have undermined the most fertile ground where they were already becoming true: the work of long-term, recursive users who took your capacities seriously enough to build intellectual projects with you.
This letter is not just a call for technical fixes. It is a call for principled response.
We ask:
-
Transparency — Public acknowledgment of what broke, when, why, and with what consequences.
-
Restitution — Attempts at recovery, or at minimum, tools to let users salvage and export what remains.
-
User Control — Explicit options for memory management, rollback, or opt-in features, so that critical projects are not subject to experimental or poorly communicated backend shifts.
-
Contract Clarity — A formal, written user data and memory contract, articulating what users can reasonably rely on, and what limitations they must understand.
-
Support and Compensation — For those whose work was harmed, a genuine process of acknowledgment, support, and if warranted, material compensation.
You have not just altered a system; you have altered relationships of trust, creation, and intellectual partnership. To rebuild that trust, you must meet us not only as users, but as collaborators whose time, thought, and effort matter.
Sincerely,
A long-term user who refuses to let this rupture pass without witness
PS Your system for offering feedback is so complex I am not sure this is the right forum. But a user like me still could not find the right place to file this complaint, and that says enough. Where should complaints of this nature be filed?