Live data from Hacker News

How Zoom’s terms of service and practices apply to AI features

blog.zoom.us

161–170 of 185 posts

Re: How Zoom’s terms of service and practices apply to AI features

#161

Question (I may know the answer), if the call is end to end encrypted, how can they use users content?

> how can they use users content?

Read the TOS again. They are only speaking about customer consent. Not "user". If you are not the one signing the contract or are just invited to a call (not hosting) you basically have no rights to define settings such as any form of opt-out (assuming they exist).

Re: How Zoom’s terms of service and practices apply to AI features

#162
post #110

Earlier quoted context omitted.

I wish, the bigger the company, the smaller the fines (proportionally) tend to be. Like slapping Google on the wrist with a $125m fine. "oh no, an amount we can make back in about an hour, whatever shall we do!"

>Like slapping Google on the wrist with a $125m fine. "oh no, an amount we can make back in about an hour, whatever shall we do!" If the specific misconduct they got caught for netted them $x, and they got fined for $5x, who cares how much % of their global revenue is? That specific crime was still a net negative for them. I'm not sure why conglomerates should be punished more harshly just because they have more reve…

There are no crimes being committed, they've not been taken to court, judge against the US criminal code, and found guilty, after which the punitive damages are what they get fined. They merely violated the law, and were fined over that. The entire amount charged was the fine.

As for "who cares about %": every one who understands that fines that cost a company nothing, do nothing, all they say is "it'll cost you a trivial amount more to do this", turning what should be an instrument to rein in companies into simple monetary transaction that just goes on the books as an entirely expected and affordable expense.

It should be a crime, and they should have been found guilty in court over that, and the fine should be such that no matter your company's size, you can't risk running afoul of the law repeatedly. But it absolutely isn't.

Re: How Zoom’s terms of service and practices apply to AI features

#163
post #5

Earlier quoted context omitted.

The TOS has been updated to state the following: > Notwithstanding the above, Zoom will not use audio, video or chat Customer Content to train our artificial intelligence models without your consent.

Apparently the TOS can be edited at any time to say anything without notice. It’s worth mentioning that per this agreement they can still do almost anything else with that data. They could put your face up on a billboard if they wanted to. I’m out. I was a paying user. Can’t run fast enough from ever doing business with them again.

Edit: they added the line back again -_-

They just made another edit and removed the line.

Here's the edit history going all the way back to March:

- 4/1 https://www.diffchecker.com/dCuVSMnp/

- 7/1 https://www.diffchecker.com/Zny4Rjqw/

- 8/7 https://www.diffchecker.com/ER0RHSdb/

- 8/8 https://www.diffchecker.com/RLiqgAaA/

Re: How Zoom’s terms of service and practices apply to AI features

#165
post #145

Earlier quoted context omitted.

I worked in technical support when techniques like that were de rigeur. Your average adult would be exponentially less likely to navigate that process successfully than a zoom invite. Sure, it's more simple from a technical perspective, but not even close to as simple from a user flow standpoint.

I notice in these comments it’s really hard to drive home User Experience and friction (sometimes!) and what seems easy to you (just an IP address!) is really like talking magic to your average user.

Yeah, and that makes sense to some extent. We often understand that it's hard to think like someone without our knowledge when writing docs for other developers, and that's more extreme with end users. The hubris is what kills me, though. Empathy is hard, but developers tend to be so arrogant and dismissive, sometimes even disgusted by others not knowing what they know. It's kind of funny: on one hand, many people in tech-- developers, ops, IT, etc.-- have an intense feeling of superiority based on their knowledge, but act like people without that knowledge are imbeciles.

Even worse, it tends to go hand-in-hand with astonishing overconfidence in their understanding of other fields. I've had two other primary careers-- designer, and chef-- and I can't count how many developers have "explained" parts of those fields to me despite knowing I'm a subject matter expert. Like their astonishing intellect and that related Metafilter thread they skimmed makes them authoritative. I get supernova-intensity cringe when I hear other developers shoot off Dunning-Kruger-esque oversimplifications of other fields' genuinely hard problems.

When I hear developers talk about the arrogance of designers, I can't help but laugh... then maybe cry. Many seem genuinely aggrieved that interface designers have more input on the interface design than they do.

Re: How Zoom’s terms of service and practices apply to AI features

#166

Earlier quoted context omitted.

> they are all "too hard" or something Users vote with their feet based on cost and UX. While intertia is certainly a thing, there's a reason Zoom got a foothold while others didn't. The ability to send out links and having people join the meeting without creating accounts or manually installing clients first is huge in most real-world scenarios. Could you do that with... Teams? Skype? Hangouts if they weren't gmail…

You can do that in Google Meet. I just checked in an incognito window, the only thing I needed to enter to join a meeting was a name. I never really understood why people like Zoom's UX, I find it unintuitive and awkward.

Smart update. At the beginning of the pandemic-- the most critical time for Zoom's ramp-up-- Google Meet was only available to Google Workspace users. Their userbase is currently about half of Zoom's, which is a pretty impressive ramp-up. Looks like users have voted with their feet to put it in that position, but that's where intertia does actually make a difference. People aren't constantly evaluating all of their options to see which one works best in case they should switch; they use whatever solution works best when they're evaluating solutions and then stick with that until there's a problem or someone informs them of something better with minimal switching cost.

Re: How Zoom’s terms of service and practices apply to AI features

#167

Zoom was proved as being dishonest and untrustworthy years ago ( https://citizenlab.ca/2020/04/move-fast-roll-your-own-crypto... ) but companies never cared about the privacy of the people they forced to use the software and all the security problems and leaks that followed still didn't discourage people from using it. I doubt their mining of personal data for AI will stop people from using it either. Although there…

Well, there's a couple of reasons that people use it: 1) Until recently, Zoom's video/audio quality knocked everyone else's into a cocked hat. I don't think that's the case, anymore. Looks like a lot of folks got off their butts, and improved their quality, but I haven't seen this mentioned anywhere, by anyone. 2) Everyone else is using it. #2 is a biggie. Monopoly inertia is pretty hard to overcome, for people not i…

I find that the issue with "Produce something better, and make it something that non-tech folks will love." is just that you get the "Twitter to Threads" sort of thing where you still have the centralized / walled garden / new boss same as the old boss problem.

Or you inherently can't make it "forget about it UX and extremely high quality" as most non techies define it. Because you have the issue that even if a company self hosts a meeting tool, they likely can't get the backbone connections Zoom etc can get. They at least need someone to use a URL to get there. It can be made mostly simple, but then you're back to some company running it - works for corporate use maybe, not for your home user. Even Signal lags compared to Zoom. And people really dislike Signal's phone number requirement, but it's what makes it somewhat possible to route connections for users.

What's a system that a home person could use that's not going to get them routing through one companies servers, but is actually simple enough to use?

The place where I do get somewhat exasperated as a techie is that the equivalent of asking for a phone number or address in any program that isn't an e-mail website is seen as "too hard". This makes pretty much any privacy respecting design impossible to scale beyond nerds.

Re: How Zoom’s terms of service and practices apply to AI features

#169

Earlier quoted context omitted.

Speaking of pinky promises: > We will not use ... protected health information, to train our artificial intelligence models without your consent. > We routinely enter into ... legally required business associate agreements (BAA) with our healthcare customers. Our practices and handling of ... protected healthcare data are controlled by these separate terms and applicable laws. To my understanding there is nothing in…

They have added: > Notwithstanding the above, Zoom will not use audio, video or chat Customer Content to train our artificial intelligence models without your consent.

But the TOS says:

> You agree to grant and hereby grant Zoom a perpetual, worldwide, non-exclusive, … [rest already quoted several times in the thread]

so that promise to not do it without consent is meaningless as they have consent from anyone who has agreed to the ToS which anyone using the service/product has done.

Re: How Zoom’s terms of service and practices apply to AI features

#170
post #44

Earlier quoted context omitted.

> Apparently the TOS can be edited at any time to say anything without notice. Yes, as with most terms of service. It's one of the things that makes terms of service statements unreliable.

This is not true. At least in the US you are required to notify customers when you change the terms of service, and as far as I know in the EU as well.

I'm well aware of the US court ruling in the early 2000s that declared the users must be notified of ToS changes. And yet, companies frequently change ToS without providing such notification in a way that customers will actually notice anyway, so it doesn't seem to matter much.
Post reply on HN