> elsewhere

In other places, LinkedIn silently opted the users into AI training. From few days ago:

https://news.ycombinator.com/item?id=41582951 - LinkedIn scraped user data for training before updating its terms of service

https://news.ycombinator.com/item?id=41584929 - LinkedIn silently opts users into generative AI data scraping by default

The FTC under Lina Khan’s enlightened, benevolent leadership seems anything but captured (though I agree with you that this is atypical): she’s the #1 enemy of everyone with a family office for a reason and if they don’t manage to throw her out, she seems to just be warming up. Everything from search monopoly to prescription drug price negotiation is on its back foot for the first time in decades.

I think people are too quick to blame regulatory capture, tbh. Today, the US only really has one world-class regulator, which sets the standard: the SEC (the FTC is more active than is used to be, but still some ways from its glory days). With the result that one of the US's two political parties is now gunning for it, and depending on the outcome of the next election it may end up neutered.

The EU is structurally difficult to effectively, ah, lobby. This, in my view, is the big differentiator; if you're a company who wants to avoid inconvenient regulations, you don't just have to compromise one political party. Unless things are very, very close, you'd have to get to, at least, tens of national parties, plus a decent bloc of MEPs, to really screw things up.

Why do you care if someone trains an AI on content you have chosen to post publicly (LinkedIn profile/posts)? I’d understand if it was your DMs or something but this stuff is no secret.

> There are some consumer protections that I really do wish we imported into the U.S., especially food safety and chemical usage.

It's always shocking when you find that some brand-name product tastes different in the EU because an ingredient used elsewhere is not considered apt for human consumption here.

After living a few years in The Netherlands, a lot of Argentinian candy has a faint taste of something kinda chemical. A few others who've moved here expressed perceiving the same flavour.

My guess so far is that living in Argentina we're continuously exposed to whatever this ingredient and become desensitised to it. When we live abroad for enough years, we become sensitive to it again.

You guys are welcome to the Cyber Resilience act if you like. which burdens software developers to risk loosing 2.5% of their turnover (Not profit, turnover) if they are unable to deliver a critical security bug to their clients years after the guarantee has expired. So you better make sure that your new startup with hardly any money gets it right the first time because every update existentially puts your company at risk.

It also means that you can't keep you margins really low as your risk remains high as it's turnover based.

If you opt out now, what's gonna happen? Do they pull out your data from the trained AI? Didn't all of this already happen, and that setting is only for your future data?

Thank you, kind Internet stranger :)

This would be a bigger issue if Linkedin cared a lot about EU users in the first place.

If tomorrow a user facing feature needed to be radically different between US and EU users, I think they'd also delay EU rollout until it reaches an actionable priority, so it's kinda par for the course.

Exactly what i was thinking.

It could be used for all sorts of things:

    You could use it to indicate the fit between employees and companies;
    You could use it to detect lies and exaggerations in CVs;
    You could use it to estimate when employees are likely to be considering seeking new opportunities.

I think it depends on what your objective is. Like if you want to simulate the LinkedIn experience, it's natural to train on LinkedIn data.

maybe someone needs an AI that generates smug self-congratulatory word salad

Training is one thing, but that feature to have AI help you with your LI posts seems psychotic. Generic generated slop for what exactly.

Exactly. Who wakes up in the morning and thinks, "You know what the world needs? An LLM that writes linkedin influencer-style posts".

HR and marketing will love it.

Make of that what you will

Well, not a general purpose one, but...

knowing LI doesn't have any NSFW content, is full of marketing content (both corporate marketing and self boasting individuals), tends to prefer positive signals (all projects succeed, etc.), is mostly done in English, and so on...

... There is clearly a market for text generation in this language bubble. Think of all the internal and external communication in your $BigCorp. That has immediate use not only in marketing, but also HR, recruitment, company policing, etc. Your next company town hall can be prepared and led (!) by this thing.

Unfortunately, the bitter news is that LinkedIn works for 98.42% of people.

Well one can't have everything.

The average webdev in many countries of Europe earns pennies and cost of living being still abysmal. Meanwhile one can only be breathing in the USA with a CS degree and easily make quite a lot, more than enough to get by.

Those "demands" for inclusive training sets are a spin attempt. And a bad one at that: It does not even address the main problem. If I refuse to be judged by a person trying to evade legal review, then that situation does not improve once they promises to make their still-hidden thought process at least as biased as the regional internet spam variety plus a few fan fiction texts for which they violated my copyrights.

The whole idea of "you cannot sue me for my discriminatory practices, because I colluded with AI company X to hide the process" is a net negative. It does not matter which side of the coin is up, when I have to pay for it.

Why can't they get that training data in another, legal way?