Key facts
- OpenAI is investigating rapid depletion of Codex credits.
- A 'warroom' has been established to address the issue.
- The company suspects fraud prevention systems are rate-limiting accounts incorrectly.
- This is causing unexpected usage drains for users.
- The investigation is ongoing.
OpenAI has initiated an internal investigation into a significant and unexpected depletion of Codex credits among its users. The artificial intelligence research company has established a dedicated 'warroom' to address the issue, which is causing credits to be consumed at a rate faster than anticipated. The primary suspicion within OpenAI is that its own fraud prevention systems might be misidentifying legitimate user activity as malicious. This misidentification could be leading to incorrect rate-limiting of accounts, thereby causing the accelerated drain of Codex credits. The situation is impacting users who utilize the Codex service for various coding and development-related tasks, leading to concerns about service availability and fair credit allocation. Further details on the specific mechanisms of the suspected fraud prevention error or the extent of user impact are not yet publicly available as the investigation is ongoing.