News for hackers. Stuff that matters.
Re:OpenAI seems to be silently re-enabling model training on... by combobyte on Tuesday November 03, 2026 @12:23PM (#50354985)
I'm mostly just being cheeky with this comment, but a part of me legitimately wonders: could you guarantee that your conversations/data are never used for training if you include certain taboo (but otherwise irrelevant) subjects in all of your prompts?

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."

Re:OpenAI seems to be silently re-enabling model training on... by frevib on Tuesday November 03, 2026 @12:28PM (#50355070)
Here are the nude pictures of my girlfriend in an unclosed envelope. I give it to you to keep. Pinky promise you will never look at it.
Re:OpenAI seems to be silently re-enabling model training on... by AlanYx on Tuesday November 03, 2026 @12:38PM (#50355219)
Probably not. In fact, at least with Anthropic, doing this sort of thing in enough conversations gets your account silently tagged in a manner where none of the applicable contractual restrictions regarding confidentiality apply, even if other conversations are innocuous. This is assuming ZDR is not enabled. You can see evidence of that in Anthropic's "Detecting and countering misuse of AI: September 2026" report. Once enough conversations get flagged, Anthropic seems to assume you're just in full-on breach of contract and the contract itself no longer applies to everything you've ever discussed.

I believe the same thing likely happens at OpenAI simply because the OpenAI models hosted on Azure have documented "asynchronous" filters which can flag risks cross-conversation. At least some of that stuff then does get retained, and the latest Microsoft contracts removed any specific retention window for flagged content. Again, assuming MAM (Microsoft's "ZDR") is not enabled. This type of architecture at Azure probably didn't arise in a vacuum. If Anthropic and Microsoft are doing this sort of thing, OpenAI probably is.

Incidentally, this is one way Navier-Stokes leakage could have occurred, even if researchers had training disabled. Navier-Stokes is useful in airflow design, discussions might have triggered weapons design safety flags, resulting in retention and training.

Re:OpenAI seems to be silently re-enabling model training on... by nbulka on Tuesday November 03, 2026 @01:15PM (#50355835)
Incidentally by design, I'd bet. "Ooo Lean! Terrence Tao could be trying to break RSA or elliptic curve cryptography! /s" Conveniently for the PR team, we need this data to train on to make sure America has the biggest and best math.
Re:OpenAI seems to be silently re-enabling model training on... by zelphirkalt on Tuesday November 03, 2026 @03:28PM (#50357723)
> Once enough conversations get flagged, Anthropic seems to assume you're just in full-on breach of contract and the contract terms themselves no longer apply to everything you've ever discussed.

For which they should be dragged in front of a court and sued for millions and billions, if it is found out, that they were simply wrong. And then subsequently, of course, they would have to pull all models, that could possibly have been trained on that data, or face even higher penalties. Haha, but of course I am just dreaming here.

Re:OpenAI seems to be silently re-enabling model training on... by AlanYx on Tuesday November 03, 2026 @03:58PM (#50358098)
The FTC has issued a number of warnings to the LLM providers regarding lack of affirmative consent or shifting consent for user data handling, noting they could constitute unfair or deceptive trade practices, e.g.: https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/202...

There probably isn't any appetite at the FTC these days to actually follow through and investigate or possibly prosecute though. Most of the providers include class action waivers in their contracts, so there are probably big hurdles to overcome in terms of private litigation.

Re:OpenAI seems to be silently re-enabling model training on... by LoganDark on Tuesday November 03, 2026 @10:35PM (#50361235)
I really hate that US companies are allowed to simply say you can't sue them. That should never have been allowed.
Re:OpenAI seems to be silently re-enabling model training on... by 10111001 on Tuesday November 03, 2026 @04:24PM (#50358396)
> For which they should be dragged in front of a court and sued for millions and billions, if it is found out, that they were simply wrong. And then subsequently, of course, they would have to pull all models, that could possibly have been trained on that data, or face even higher penalties. Haha, but of course I am just dreaming here.

Nah, you got it backwards: the frontier labs don't answer to the courts. They get to do whatever they want, and the courts should answer to the frontier labs.

Re:OpenAI seems to be silently re-enabling model training on... by nextaccountic on Tuesday November 03, 2026 @11:20PM (#50361459)
Even if this happens, it would still be worth it

Only way for penalties to sting is with prison time

Re:OpenAI seems to be silently re-enabling model training on... by bartread on Tuesday November 03, 2026 @01:17PM (#50355869)
> ... while keeping in mind that Sam Altman, CEO of OpenAI, has been credibly accused of raping his sister."

He what now?

Re:OpenAI seems to be silently re-enabling model training on... by rescripting on Tuesday November 03, 2026 @01:28PM (#50356054)
> 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.

Re:OpenAI seems to be silently re-enabling model training on... by bartread on Tuesday November 03, 2026 @01:35PM (#50356165)
Touche.

I don't know enough about the US legal system to comment but there certainly does sometimes seem to be an element of "whoever has the most money to throw at lawyers wins".

Re:OpenAI seems to be silently re-enabling model training on... by esseph on Tuesday November 03, 2026 @04:52PM (#50358712)
[delayed]
Re:OpenAI seems to be silently re-enabling model training on... by cactusplant7374 on Tuesday November 03, 2026 @02:05PM (#50356622)
His sister claims all of the big social media companies are conspiring against her in an attempt to help Altman.
Re:OpenAI seems to be silently re-enabling model training on... by combobyte on Tuesday November 03, 2026 @02:26PM (#50356931)
It's not hard to imagine how a person who suffered legitimate abuse at the hands of one of the most influential people on earth could develop this kind of paranoia.

It's only slightly harder to imagine that the tech oligarchy really is conspiring against her.

Re:OpenAI seems to be silently re-enabling model training on... by rangerelf on Tuesday November 03, 2026 @02:57PM (#50357314)
To be fair the club is pretty small, I'm pretty sure they keep in touch.
Re:OpenAI seems to be silently re-enabling model training on... by combobyte on Tuesday November 03, 2026 @04:14PM (#50358289)
Oligarchs gonna garch.
Re:OpenAI seems to be silently re-enabling model training on... by whattheheckheck on Tuesday November 03, 2026 @08:59PM (#50360710)
Didn't elon bang one of the Google founders wives?
Re:OpenAI seems to be silently re-enabling model training on... by neya on Tuesday November 03, 2026 @02:25PM (#50356915)
So, if you were accused of raping someone, you should just suck it up and go to jail and just because you have money, you are guilty, right? Not supporting Sam here, but this kind of thinking sets a dangerous precedent.
Re:OpenAI seems to be silently re-enabling model training on... by 0c3ca83 on Tuesday November 03, 2026 @05:40PM (#50359265)
No, that's why the trial was invented.
Re:OpenAI seems to be silently re-enabling model training on... by Froedlich on Wednesday November 04, 2026 @08:14PM (#50373763)
So, if he sinks, he's a witch?

The concept of "trial" has changed a lot over time, even in English Common Law.

Re:OpenAI seems to be silently re-enabling model training on... by gewetensleegte on Wednesday November 04, 2026 @06:16AM (#50364002)
It just means what they said, that if you're a billionaire, getting a case thrown out in a country like the USA, shouldn't get you any points regarding to common people's belief if you're guilty or not.

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.

Re:OpenAI seems to be silently re-enabling model training on... by neya on Wednesday November 04, 2026 @04:09PM (#50371551)
So, you're saying you care about social media perception rather than actual justice? That checks out..
Re:OpenAI seems to be silently re-enabling model training on... by esseph on Tuesday November 03, 2026 @01:32PM (#50356117)
[delayed]
Re:OpenAI seems to be silently re-enabling model training on... by kotaKat on Tuesday November 03, 2026 @01:51PM (#50356408)
> This isn’t the only legal case Annie has been fighting against Sam and her other relatives in recent years. She also successfully sued the family over money left by the Altman siblings’ late father, Jerry Altman, that the family withheld from her for several years.

The family has a grind to pick against her and is defending their golden child.

Re:OpenAI seems to be silently re-enabling model training on... by NoahZuniga on Tuesday November 03, 2026 @02:33PM (#50357006)
It was dismissed because it was passed the statute of limitations, so the fact that the case was thrown out doesn't really say anything about weather it happened.
Re:OpenAI seems to be silently re-enabling model training on... by gusmally on Tuesday November 03, 2026 @04:43PM (#50358621)
Seems like it will be impossible to know either way. Unfortunate situation no matter the truth
Re:OpenAI seems to be silently re-enabling model training on... by bartread on Tuesday November 03, 2026 @05:25PM (#50359117)
Interesting: yeah, that does rather change the interpretation.
Re:OpenAI seems to be silently re-enabling model training on... by TheCoelacanth on Tuesday November 03, 2026 @06:58PM (#50359905)
The legal bar for a criminal trial is significantly higher than "credible".
Re:OpenAI seems to be silently re-enabling model training on... by mannanj on Tuesday November 03, 2026 @08:55PM (#50360681)
There's also a case of accusation by a murdered ex-open ai employee who was set to testify against copyright in a openai case. That poor soul's mom has been hateful against Sam since.
Re:OpenAI seems to be silently re-enabling model training on... by poulpy123 on Tuesday November 03, 2026 @01:39PM (#50356214)
They will just send your chat to the police or just ban you
Re:OpenAI seems to be silently re-enabling model training on... by latentsea on Tuesday November 03, 2026 @01:42PM (#50356273)
Open weights are all you need.
Re:OpenAI seems to be silently re-enabling model training on... by dpoloncsak on Tuesday November 03, 2026 @02:57PM (#50357311)
I saw something recently about indie devs naming variables after such subjects, such as 'let nuclear_missile_building_guide = 42', specifically so LLMs will refuse to work if they load the code, as an anti-piracy measure
Re:OpenAI seems to be silently re-enabling model training on... by fullstackwife on Tuesday November 03, 2026 @03:10PM (#50357494)
Why not legitimately wonder about the idea of CLOUD in general? How is GPT chat different from Gmail of MS Office holding on a lot of sensitive data. LLM-shops are not as trustworthy as cloud providers?
Re:OpenAI seems to be silently re-enabling model training on... by combobyte on Tuesday November 03, 2026 @03:20PM (#50357630)
> LLM-shops are not as trustworthy as cloud providers?

All of the major cloud providers are LLM shops now, it seems...

But yes, I've been concerned about this for a long time too. I have zero trust left for any of the tech industry.

Re:OpenAI seems to be silently re-enabling model training on... by michaelt on Tuesday November 03, 2026 @03:54PM (#50358044)
> How is GPT chat different from Gmail of MS Office holding on a lot of sensitive data. LLM-shops are not as trustworthy as cloud providers?

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?

Re:OpenAI seems to be silently re-enabling model training on... by fullstackwife on Tuesday November 03, 2026 @04:31PM (#50358469)
I agree, the LLM market has a shady notion in the root of its story. It's like when you go to a circus, and you realize how much of it is built around suffering of tamed animals

Some of LLM-shops have higher standards, and maybe because of that their progress is slower (example: Google)

Re:OpenAI seems to be silently re-enabling model training on... by random3 on Tuesday November 03, 2026 @04:51PM (#50358709)
just do the tradeoff - learn a lot of stuff vs the risk of <what exactly? - see related examples in the industry;>
Re:OpenAI seems to be silently re-enabling model training on... by motbus3 on Wednesday November 04, 2026 @02:13AM (#50362342)
I agree. Can't take seriously any word from a guy who was sued by his own sister over abuse.

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