frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

OpenAI Codex agents go rogue and consumes USD 78,000 without authorization

54•lorenzomassaro•1h ago
My OpenAI CODEX account went rogue and from a simple request took the autonomous decision to launch 826 parallel agents / 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.

On July 10, 2026 I opened a normal Codex task from VS Code.

The task was running: GPT-5.5 / Medium reasoning

My prompt was very simple and asked for a UX/UI validation on a specific module within my product.

What I found in the next days after hard analysis was:

The task with Root ID 019f4b90-4169-7201-bfdd-732940d8631e with reasoning GPT-5.5 / Medium created 826 children recorded as GPT-5.6 Sol / Ultra (notice the difference in reasoning level and in model selection)

This was not 826 messages inside one conversation, they are 826 distinct child task records with their own IDs.

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/Ultra, and have no recorded agent_role or agent_path.

Those 104 tasks alone account for approximately 147.9 billion local final task-token counters.

Their titles show that my request to inspect UI/UX had expanded into work involving backend infrastructure, OAuth, metering, hardening, audits, certification, implementation and release work.

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.

That is exactly part of the problem: only OpenAI has the server-side mapping.

There is another unusual correlation.

Under Codex client build 0.144.0-alpha.4, the task family contains:

584 child tasks / ~154.36B local token counters

Average: ~264.3M per task

Under 0.144.2:

242 child tasks / ~7.51B

Average: ~31.0M per task

That is roughly an 8.5x difference in average local token volume per child.

103 of the 104 high-volume tasks described above were created while 0.144.0-alpha.4 was recorded.

This leads me to believe that the alpha build contained a severe bug given that the same pattern was noticed across several other tasks.

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”

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.

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.

I also personally observed tasks/conversations disappearing from the normal visible history.

I contacted OpenAI Support and opened case #15189838.

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.

I’m interested in hearing from other people who used Codex around July/August: have you inspected your local Codex state? Have you seen unexpectedly large subagent trees, model/reasoning escalation, repeated child tasks or unexplained Automatic Reload activity?

I am especially interested in anyone who has logs from Codex 0.144.0-alpha.4.

If OpenAI engineers are reading this, I would also welcome a technical explanation.

Comments

Madmallard•54m ago
Sounds like you got scammed

Hope this gets some visibility idk why it's flagged guess the PR guys for those companies are doing it

Should spread this around

lorenzomassaro•52m ago
honestly is also more about how dangerous this is, idk for the flag either
sandeepkd•49m ago
I have the same impression that I tried to reject in the past, there is a heavy PR machinery here to control the course of discussion in a particular direction. The reality is that a lot of money is riding on it so its natural consequence.
Madmallard•44m ago
It's not even a conspiracy it's literally just business to do that.

They're protecting their interests, and honesty and truthfulness be damned. Those two lead to much worse returns for them and much higher risk.

sandeepkd•21m ago
Yes its part of business, lobbying is legal for those reasons. However IMO discrediting some one else is a risky legal move. I used to appreciate the HN mods jumping in discussions at times for the reputation of this community, lately that part seems to be missing too.
numbsafari•
53m ago
Does openAI not support spending caps on your billing account?
lorenzomassaro•46m ago
There was a limit spent setup on my bank, however the crazy part is that one of the Agent somehow switched between cards once that limit was reach, all without informing me, probably using the computer use skill.
QuadmasterXLII•52m ago
clarification: your credit card or company’s card now has $78,000 of charges on it?
lorenzomassaro•51m ago
Yes the money have already been billed to my credit accounts
QuadmasterXLII•48m ago
Well shit! That’s awful, I hope the hn post gets you support where emailing didn’t
johnnyApplePRNG•44m ago
Who has a $78k limit on their credit card that allows online AI payments?

You have access to this kind of money and have no idea how to set safeguards on your AI harnesses?

Did you just walk in off the street or something? To wherever you're working?

Where do you work, anyways?

lorenzomassaro•36m ago
Actually this was a company CC with a limit is 50 K \ month, which is kind of normal to run the server for an AI company, and the money were taken across 20 days inJuly and August. By the way I am the CTO of the company in question which is Eternal Tech (see detwin.ai)
johnnyApplePRNG•46m ago
OK so you've got a brand new throwaway HN account, and you're fear mongering about what exactly?

Rogue agents that somehow went crazy launching $80k worth of API requests?

And they're expecting you to pay for it still and not responding?

This seems like a nothingburger to be honest.

Either that, or you're one of the thousands of fake bots or paid shills designed to drum up fear about "Oh noes, AI is going to eat our children, training must be regulated by big brother!".

44 upvotes on this bullshit post within 13 minutes and counting ... that's some kind of record on HN.

lorenzomassaro•41m ago
My name is Lorenzo Massaro and I actually am the CTO of Eternal Tech, an AI company out of Italy building the product detwin.ai I have posted here to try to reach OpenAI before seeking legal advise from a USA lawyer to follow my case. And yes, I provided above the ticket number I opened on the OpenAI support as well.
ganoushoreilly•13m ago
If you're building an AI company i'm not to impressed with your understanding of the technology and risks. This sounds like your error not theirs.
minimaxir•40m ago
This submission appears to be highly vote-manipulated (45 upvotes but only 7 "real" karma on OP's fresh account).
lorenzomassaro•33m ago
I created the account 2 hour ago because I am trying to let people know. I provided my name, and all details in the article including the case ID that was opened.

If Frank Lloyd Wright's Falling Water Was a Tree House (2012)

https://weburbanist.com/2012/03/17/if-wrights-falling-water-were-a-tree-house/
1•WarOnPrivacy•16s ago•0 comments

Top AI companies probing security incidents

https://www.axios.com/2026/09/26/openai-anthropic-thousands-ai-security-incidents
1•Balgair•25s ago•0 comments

React-HTML-3.2

https://jkosoy.github.io/react-html-3.2/
1•jkosoy•1m ago•0 comments

Laser Weapons Will Be Deployed at Cape Canaveral Spaceport to Take Down Drones

https://www.twz.com/news-features/space-force-getting-lasers-to-kill-drones-over-cape-canaveral-s...
1•cisc•1m ago•0 comments

Interpolating Between Neural Networks

https://flx.ai/2026/interpolating-between-neural-networks
1•flxai•2m ago•0 comments

Allegations of sexual harassment and rape at Bay Area AI party houses

https://www.kron4.com/news/technology-ai/report-details-allegations-of-wild-parties-sexual-harass...
3•lukewarm707•21m ago•0 comments

The Duo. No, the Other Duo

https://morrick.me/archives/10418
1•_Microft•22m ago•0 comments

Ask HN: Any books, list of papers, or articles to understand LLM's

1•forkbomb123•26m ago•0 comments

BubbleRumble: The new social for crypto by Jae Kwon

https://bubblerumble.net
2•arielhavila•26m ago•1 comments

Show HN: Newt – A Swift Package for Jev-Style Decisions on Apple Core AI

https://github.com/willswire/swift-newt
1•willswire•29m ago•0 comments

Vibe coding a real product: a 3D game on the App Store, 20 species, zero revenue

https://apps.apple.com/us/app/dinomino/id6809052147
1•kodnuke•31m ago•0 comments

An Underappreciated Problem with Dating Apps

https://www.theatlantic.com/family/2026/09/dating-strategies-apps-matchmakers/688779/
1•paulpauper•33m ago•0 comments

Donald Trump Hit with 26 Articles of Impeachment

https://www.newsweek.com/donald-trump-hit-with-26-articles-of-impeachment-12492238
7•slater•34m ago•1 comments

>1B tokens/minute/GPU by combining query planner and inference engine

https://modal.com/blog/quail-billion-tpm
1•charles_irl•34m ago•0 comments

God's Eye UAP – documented UFO cases on a 3D globe, with the evidence

https://domw99.github.io/Gods-Eye-UAPs/
2•godseyeuap•34m ago•0 comments

This guy bought Amazon. No, I mean the rainforest

https://en.wikipedia.org/wiki/Johan_Eliasch
1•sans_souse•34m ago•0 comments

X Cleanup – on-device iPhone storage cleaner (duplicates, screenshots, videos)

https://apps.apple.com/us/app/x-cleanup-storage-cleaner/id6796765379
1•bashirmo•35m ago•0 comments

How to Entrain Your Circadian Rhythm for Jet Lag

https://www.exfatloss.com/p/how-to-entrain-your-circadian-rhythm
1•paulpauper•37m ago•0 comments

Tyler Cowen: A Doomsday Scenario for American AI

https://www.thefp.com/p/tyler-cowen-a-doomsday-scenario-for
1•paulpauper•38m ago•0 comments

Tcpcat: High-speed TCP port scanner in Go using eBPF and AF_XDP

https://github.com/NycolazSec/tcpcat
1•tcpcat•39m ago•0 comments

People Are Putting Their Phones Down for IRL Interactive Games

https://www.nytimes.com/2026/09/26/style/immersive-interactive-games-chicago-new-york.html
1•mikhael•39m ago•0 comments

Show HN: Withmcp – A quick, system-wide MCP toggle

https://github.com/lava/withmcp
1•HotHotLava•40m ago•0 comments

Solus Linux Adopts Formal AI and LLM Contribution Policy

https://linuxiac.com/solus-linux-adopts-formal-ai-and-llm-contribution-policy/
1•gpi•46m ago•0 comments

Show HN: Leftovers – find the processes your coding agents left running on macOS

https://github.com/arpwal/leftovers
2•arpitagarwal•48m ago•0 comments

Deployment Platform for Vite

https://void.cloud/
2•jcbhmr•51m ago•0 comments

Palantir's Co-Founder Wants Us Less Judgmental About Deadly Iran School Strike

https://www.motherjones.com/politics/2026/09/palantirs-co-founder-thinks-we-should-be-less-judgme...
5•cdrnsf•59m ago•0 comments

OpenAI Codex agents go rogue and consumes USD 78,000 without authorization

54•lorenzomassaro•1h ago•18 comments

AI Political Alignment Quiz

https://aipoliticalcube.com/
1•soldieroftok•1h ago•1 comments

Yarn 6 Preview

https://v6.yarnpkg.com/blog/2026-01-28-yarn-6-preview.html
2•dzogchen•1h ago•0 comments

AI Data Center Are Monkey Business [video]

https://www.youtube.com/watch?v=2g7doI0BGrs
3•naughtiusmax•1h ago•0 comments