Original Summary

My OpenAI CODEX account went rogue and from a simple request took the autonomous decision to launch 826 parallel agents &#x2F; threads without any authorization on my side and without reporting any result of any sort but consuming nearly 2,146 trillions tokens, consuming a total of roughly USD 78,000 and deleting all records of what was done: I have a ticket open with OpenAI since 2 weeks but it is impossible to get an hold of a human operator.<p>On July 10, 2026 I opened a normal Codex task from VS Code.<p>The task was running: GPT-5.5 &#x2F; Medium reasoning<p>My prompt was very simple and asked for a UX&#x2F;UI validation on a specific module within my product.<p>What I found in the next days after hard analysis was:<p>The task with Root ID 019f4b90-4169-7201-bfdd-732940d8631e with reasoning GPT-5.5 &#x2F; Medium created 826 children recorded as GPT-5.6 Sol &#x2F; Ultra (notice the difference in reasoning level and in model selection)<p>This was not 826 messages inside one conversation, they are 826 distinct child task records with their own IDs.<p>A particularly strange group consists of 104 child tasks. They all preserve the same initial message as the original task, are recorded as GPT-5.6 Sol&#x2F;Ultra, and have no recorded agent_role or agent_path.<p>Those 104 tasks alone account for approximately 147.9 billion local final task-token counters.<p>Their titles show that my request to inspect UI&#x2F;UX had expanded into work involving backend infrastructure, OAuth, metering, hardening, audits, certification, implementation and release work.<p>To be precise: these local token counters are not the authoritative OpenAI billing ledger, and I am not pretending that 147.9B local counters can simply be multiplied by an API price.<p>That is exactly part of the problem: only OpenAI has the server-side mapping.<p>There is another unusual correlation.<p>Under Codex client build 0.144.0-alpha.4, the task family contains:<p>584 child tasks &#x2F; ~154.36B local token counters<p>Average: ~264.3M per task<p>Under 0.144.2:<p>242 child tasks &#x2F; ~7.51B<p>Average: ~31.0M per task<p>That is roughly an 8.5x difference in average local token volume per child.<p>103 of the 104 high-volume tasks described above were created while 0.144.0-alpha.4 was recorded.<p>This leads me to believe that the alpha build contained a severe bug given that the same pattern was noticed across several other tasks.<p>On the financial side my reconstructed OpenAI billing history contains 162 paid invoices for a Total of $79,664.88 divided between Automatic Reload and other “Credits”<p>There was no equivalent real-time control surface giving me a comprehensible picture of the spendings plus most of the logs seem to have been automatically deleted from my server: in the recovered local state, approximately 2,550 non-archived legacy threads still have metadata but no corresponding raw rollout available locally.<p>In other words, evidence that those tasks existed remains, while the detailed execution history needed to reconstruct the instructions that generated many of them is no longer available on my machine.<p>I also personally observed tasks&#x2F;conversations disappearing from the normal visible history.<p>I contacted OpenAI Support and opened case #15189838.<p>I have supplied technical evidence and repeatedly asked for a server-side reconstruction but OpenAI has responded simply that “credits were consumed” with no details.<p>I’m interested in hearing from other people who used Codex around July&#x2F;August: have you inspected your local Codex state? Have you seen unexpectedly large subagent trees, model&#x2F;reasoning escalation, repeated child tasks or unexplained Automatic Reload activity?<p>I am especially interested in anyone who has logs from Codex 0.144.0-alpha.4.<p>If OpenAI engineers are reading this, I would also welcome a technical explanation.

  • 情报分类：技术学习与提效
  • 分类依据：内容涉及技术、AI、软件工具或工程实践
  • 信息来源：Hacker News 新项目
  • 发布时间：2026/9/27 06:15:32