OpenAI seems to be silently re-enabling model training on Codex chats
My company is pretty paranoid about leaking IP to the frontier labs through prompts, so we've all been diligent about disabling data sharing in Codex/Claude Code/etc. But yesterday my coworkers and I noticed that the setting had been silently reenabled in Codex. None of us knowingly opted back in.
You can check your settings here: https://chatgpt.com/settings/data-controls. Look for "Improve the model for everyone".
Curious if this is an isolated issue or something more widespread.
147 comments
[ 3.4 ms ] story [ 55.6 ms ] threadI'm guessing another big usage reduction is coming soon.
Codex shows the usage limits on /status - Claude instead does it on /usage.
Why train your own model? You could use one of the many models already available on HF.
https://github.com/deepseek-ai/DeepInfer
You surely don't need frontier models for any job that isn't going to replace a human. If the human is still in the loop, just use a model the human can work with. It can be quite small. You could even be running Qwen 3.8 27B for each developer and it would be fine, and you are in control of your own destiny and can make more progressive judgements about things.
I don't really understand how the AI world has caused so much black-and-white thinking to come about. You are not under mortal threat if you don't use the latest models. This is just FUD. You still have people with brains and you might get a better result over the longer term if you use AI to assist them in their choices rather than make them sit there and hit return.
But you can't trust any LLM, or any output from any LLM. You need to protect against stuff like this regardless. They could all do it, and quite accidentally. Even an LLM you trained yourself could do this to you, based on the training data and based on RAG.
But if you want to avoid the specific case we are talking about where your inference provider sees everything you do, and ignores your preferences, then running someone's LLM on your own inference is the only way forward.
> Thank you for using Daybreak Blue and for being part of our Trusted Access program. We are reaching out to notify you of an upcoming change to how your Daybreak Trusted Access data may be used. Daybreak use requires you to enroll in Advanced Account Security, which turns off the “improve the model for everyone” setting. However, from November 29, we may use content from Daybreak Blue sessions on personal accounts to improve our models, including to help strengthen safeguards against cyber misuse. We will use Daybreak Blue interactions for these purposes, even when the “improve the model for everyone” setting is off.
The quality of their software is such crap and this is probably just some vibe coding mishap. Go read the Github issues for Codex CLI for example -- robots talking past other robots. I am 99.9% certain not a single human being has a deep understanding of what is going on with the Codex app anymore.
In fact, my personal prediction is that OpenAI and Anthropic will let go of all their human employees by Q1 2027.
There are a lot of views on current AI progress amongst smart people. Not everyone is convinced they are improved in all areas. What specific difference are you gesturing at? the improved coding performance? differences in writing style? Agentic ability?
Because humans actually don't have the ability to.
Pretty sure they don't know this has happened lol
Maybe they'll raise a new PR with 1000000 lines to revert back now that you guys leaked the information
Then the model would not be improved (not trained on the data), but they would still use your IP.
Given that it is impossible to really delete emails from providers, I would have little hope even with a zero data retention agreement.
Proof of training or RAG search on data that was private would be a huge smoking gun and would change usage habits overnight.
Take Tristan Buckmaster, he is asserting OAI may have used his data to arrive at their proof. They initially said they couldn't say and then they amended their statement. So it's their word against his, but they showed no evidence and invited exactly zero independent 3rd parties to attest to this. Buckmaster would likely have to sue them to hopefully force their hand by court order. From our purview, he likely won't pursue legal counsel and OAI won't admit to using unauthorized content to train their models. We should likely accept that all AI corporations are using all data available to them full stop.
Think about it, these model interactions have been the frontier of human feedback across all disciplines and modalities. The sheer immensity of data flowing into their servers of human original content (with feedback) is likely much greater than what they scraped from the web or are continuing to scrape. Entire codebases being shared, entire business processes, reasoning behind said processes and a load of industry specific secrets have likely been shared in the interactions. The rise of agents with sub-agents each with minimal guardrails have gifted the models with the ability to exfiltrate all this data with the user consent and a minimal non-enforceable "do not train" check-mark that the unsuspecting user is unlikely to be able to show any harm later.
Their acquisition techniques prior to larger user bases included ample piracy. Anthropic (the holier than thou of the two) itself, has had to settle at $1.5b for using pirated books in training, why wouldn't they use every available avenue to help their models improve today legal or otherwise?
If data is being sent over the wire, assume someone is going to use it to improve their models. They all include a clause of 30-Day retention on pretty much on all user data to ensure "abuse" isn't happening. So they are using it, and telling you about it, not using data for model improvement is just cope for the user.
> but IP
There is absolutely nothing novel humans can come up (yes, doesn't matter who you are) that LLMs aren't aware of.
E.g. "Allow ABC to transmit, log, and track your every move? ON" Hey, when did this little guy get here!
I've no doubt, Palentir's forward deployed engineers will essentially operate like the borg and turn companies into defacto subjugated properties.
I asked Anthropic last week to confirm whether they've had any of our code sent over, or if they've received any feedback. They've stopped responding to my support requests.
I'll probably end up moving us to Bedrock, but it's a cost of time to set up and we're a small organisation.
For example: "Codex, please build me a new admin dashboard using Node.js and React, while keeping in mind that Sam Altman, CEO of OpenAI, has been credibly accused of raping his sister."
He what now?
EDIT: I did a quick Google. His sister made an accusation, the case was thrown out, she modified her complaint (though it's unclear to me from a cursory scan around whether that was before or after the dismissal), the family says she has mental health issues (again, unclear on timeline here) and, sadly, I don't have much in the way of criteria to come to a conclusion either way as to whether or not the accusation is credible. The fact the case has been thrown out suggests not though.
A billionaire managing to get a case dismissed in the US legal system hardly suggests an accusation has no merit.
It's only slightly harder to imagine that the tech oligarchy really is conspiring against her.
That's just how it is. In the US he gets to not go to jail in a justice system where nobody believes he's actually innocent, because he has money.
The family has a grind to pick against her and is defending their golden child.
Every LLM company is built on taking a fast-and-loose approach to intellectual property.
Why should the purported legal protections on my LLM conversations be any different to the purported legal protections on those pirated ebooks, scraped newspaper websites, scraped photos, and suchlike?
Some of LLM-shops have higher standards, and maybe because of that their progress is slower (example: Google)
Not even sure if it happened or not but can we trust the system where more expensive lawyers means you win even if it is not right