The ChatGPT Android app still sometimes locks one CPU core at 100% even when idle. It often lags even with no chat loaded; projects temporarily disappear; voice messages vanish; and message history occasionally breaks after edits (only “Clear storage + relogin” helps).

These issues have been reported for more than a year across multiple threads and platforms (CPU drain, message loss, sync failures, voice bugs) — yet no clear fix ever lands.

This isn't only about a buggy app. It shows what happens when a company optimizes entirely for model research and growth, while the interface — the literal body through which people experience AI — decays in silence. Most Android feedback channels are closed, community threads get archived, and developers can't even inspect the client to help.

If the app were open-source, engineers outside OpenAI could trace the leaks, profile CPU behavior, and contribute patches — the kind of cooperative debugging that open ecosystems are built on.

It would free OpenAI’s core team to focus on new LLMs and the impact of AGI on society instead of endless patches, and users would finally get a reliable interface.

Everyone wins.

If this discussion gains enough visibility, someone at OpenAI will potentially decide to let the community help.