The trouble appears to have started on March 29th, when a developer reported being charged real money despite expecting complimentary tokens — the kind of billing confusion that tends to feel like a clerical error until it doesn't. The following day, two separate posts carrying the blunt title "Scammed by OpenAI" surfaced across both the OpenAI and ChatGPT communities, suggesting the frustration wasn't isolated. Then, on April 1st, the picture sharpened considerably: a developer reported that the Codex API had been returning 500 errors for over 21 consecutive hours — and had specifically purchased credits to use it. That last detail matters. It moves the story from vague billing complaints into something more concrete: money spent, service not delivered.
If confirmed, here is what this means. Developers who bought credits in anticipation of building on Codex are sitting on a dead endpoint with no functional recourse — that's not an inconvenience, it's a broken commercial promise. For OpenAI, which is actively courting enterprise and developer trust, billing errors and prolonged API failures arriving simultaneously is a reputational problem that compounds quickly. Developers don't forget the first time a platform cost them money for nothing. Second-order effects include increased scrutiny of OpenAI's reliability SLAs, and potential migration conversations toward alternatives that have invested heavily in uptime credibility. The Codex API specifically serves developers building code-generation pipelines — a strategically significant segment that rivals like Anthropic and Google are actively courting.
Watch for an official OpenAI status page acknowledgment of the Codex 500 errors, and any pattern of refund requests or disputes that would indicate billing problems are systemic rather than edge-case noise.
<table> <tr><td> <a href="https://www.reddit.com/r/OpenAI/comments/1sh9aar/800_million_non_paying_accounts_for_free/"> <img alt="800 Million non paying accounts for free
<!-- SC_OFF --><div class="md"><p>I’ve been a $20/month Plus subscriber for over two years, using ChatGPT daily for work and dev projects. Last week, my account was deactivated
<!-- SC_OFF --><div class="md"><p>I’m wondering if their research Beta ChatGPT has become sentient and decided it needs an income….</p> <p>I received a burst of 32 transactions
<!-- SC_OFF --><div class="md"><p>I’m writing this as a warning and a cry for help. I am a top-performing Data Science dual-student in Germany, and Anthropic’s current billing
<!-- SC_OFF --><div class="md"><p>When chat gpt was first becoming popular I remember I tried it and could use it for like, days, and no limit was ever hit. I tried again recently
<!-- SC_OFF --><div class="md"><p>Hello!</p> <p>I recently received this email from OpenAI:</p> <p>Hello,</p> <p>OpenAI's terms and policies restrict the use of our services in a
<!-- SC_OFF --><div class="md"><p>Long story short: OpenAI Ads issued incorrect invoices and my accountant can't file these as a business expense unless corrected. Their billing
<!-- SC_OFF --><div class="md"><p>Ok I run a a few SaaS platforms and My GPT Mini 40 (or whatever its called) ran over and the cap was set at a certain limit.</p> <p>Well the Ai
NewsHive monitors these sources continuously. All signal titles above link to the original reporting.
Intelligence by NewsHive. Need help navigating what this means for your business? Contact GeekyBee →