Hacker News Reader

Comment by thih9 | Hacker News Reader

Context:

- “Meta has a new stand-alone AI app. It lets you see what other people are asking.” https://www.businessinsider.com/meta-ai-app-public-feed-warn... (may 2025)

- “People are seemingly accidentally publishing their AI chat histories with Meta’s new AI app” https://www.crikey.com.au/2025/05/05/meta-ai-chatbot-discove... (may 2025)

Note that the first link states that conversations are private by default and that user error is likely involved[1]. Mozilla’s use of text emphasis almost implies otherwise[2].

[1]: “To be clear, your AI chats are not public by default — you have to choose to share them individually by tapping a share button. Even so, I get the sense that some people don't really understand what they're sharing, or what's going on.”

[2]: At least that’s how I understood “_Make all AI interactions private by default_ with no public sharing option unless explicitly enabled through informed consent.” at first glance.

View on HN

I'm not sure about this specific situation, but from Google Docs to ChatGPT to Notion, there's a clear distinction between "make this a shareable link to only those who have the link" and "also make that shareable link searchable/discoverable by the public."

If Meta is turning that "searchability/discoverability" on by default when a share button is activated on an AI chat - or worse, if they're not even giving this industry-standard option - that would both explain the confusion, and be a terribly unexpected dark pattern. As the parent notes, the activation of a share icon is not informed consent.

I don't think that is the case with ChatGPT: https://www.vice.com/en/article/chatgpt-users-report-being-a...

That situation was a bug, with a detailed post-mortem: https://openai.com/index/march-20-chatgpt-outage/

It is worth noting of course that per https://news.ycombinator.com/item?id=44185913 ChatGPT will now be required by law to retain a record of all chat history for courts and lawyers to review, but at least that data won't be made fully public.

Both of those are different from a willing and intentional dark pattern of making things publicly discoverable!

I did a test in the app, and it is pretty obvious you are posting the chat. You click share, then you are given a preview, and you have to hit post to actually post it publicly.

There’s a chance you’re overestimating a large percentage of the public.

Then should publishing always be disallowed on all platforms? I’m having trouble understanding how is what Facebook is doing any different from ChatGPT, and in general all web apps.

ChatGPT sharing is anonymous + you are also having to share the link itself, it doesn't randomly go to other people's feeds.

Oh that’s terrible, I didn’t get that part.

If a part of the population is dumb, the issue is that it’s dumb, not meta user interface.

I’m the first to bash on meta but there are things that they are not responsible for.

No, the issue is that Meta is weaponising that stupidity by breaching naive trust, and that's... stupid too. But for different reasons.

If you don't communicate at a level suitable for your audience, then you're not communicating. At best, you're just making noise. At worst, you're being a manipulative sack of shit. I do not believe Meta are in the noise-making business.

Life is tough, but it’s tougher if you’re stupid.

And then does it ask who you want to post it to, or to which app? Because that's a fairly common pattern when I "share" something on my phone.

Or does it just post it onto the public feed?

I haven't used the app and likely won't but neither article shows the mechanism by which users opt into "sharing" their interactions. Is there a dark pattern involved?

Like, in lots of other app contexts, you can hit "share" and then get a modal that gives you options (do you want to share it via WhatsApp or messages or email or ...) and only after you select a mode and a recipient does it actually get shared -- but if you make a "share" button whose behavior is "immediately publish", people might reasonably be surprised if they actually just want to share the results with a specific trusted person who they expected to select next in the interaction.

Thank you for this. One of these links should really be the one in the submission. The Mozilla petition doesn't provide any useful context.

Thank you, the linked article was completely unclear and vague on what the issue was.

Seems odd to me, I use the app and it has never once nudged me to share anything?

Perhaps you're sharing things without knowing it.

I did a test to see how it works, the app make it pretty clear. You click share, it then shows you a preview of exactly what you are posting, and you have to hit a post button to actually post it.

You are a power user. How about the median meta user? Would they know? Is the app designed to alert them?

Are you pointing out an actual issue with this app? Or are your questions just hypotheticals about some mythical app that Meta may or may not have built?

Has someone noticed a similar thing with ChatGPT and private Github repos? ChatGPT has recommended private repo links to me many times. Because they are not public, I get repo not found error. But ChatGPT can generate private code with no issues.

Are you sure the private repos exist, rather than just being hallucinations?

Yeah it's always been hallucinated repos. It's easy enough to verify - do a google search on the repo name itself and see if it's ever been mentioned or used anywhere and if it matches what ChatGPT thinks it was suggesting. And check to see if the owner of the repo exists and if they do, check their repositories to see which programming languages & topics they favor and see if it matches or not.

Github -> Microsoft sharing everything with OpenAI sadly sounds realistic.

Or they don't exist in the first pLace, or they're now deleted repos, or they used to be public but went private, etc.

Unfortunately, everything you mentioned above is a possibility.

Honestly, even after (quickly) reading the context, I still don't know why Mozilla had such a strong reaction to this.

Maybe one of their execs fat fingered and shared something embarrassingly.

That context should be in there. Give me the facts or else it’s just whining.

Sidenote: How is Mozilla's comm's so bad? I had no idea what their petition was talking about.

Thank you for your service.

It's honestly impressive how bad Mozilla is at communication and its been like this for years.

Effective communication is quite humbling.

It's been my experience that I need to take full responsibility for the effectiveness of my communications.

A few years ago, I threw together a PowerPoint show, based on Randall Munroe's Communication comic[0]. I did it for an organization I participate in, that is full of some of the worst communicators I've ever encountered.

It astounds me, how people that get paid to communicate, don't understand the fundamentals.

[0] https://docs.google.com/presentation/d/1qQDAuhGvBvBlZVH2zn_V... (You need to view the slide notes, or it doesn't make any sense).

Anyone know how to show the speaker notes on mobile?

It seems like the xkcd comic itself is pretty hard to follow. It would probably be better with dialog.

I'm not sure. It was originally a PP show, and I dumped it into Google Slides, for a friend.

Probably doesn't matter. My experience is that most folks have no intentions of doing what it prescribes, so it's kind of wasted effort.

Not what you asked for but should help https://www.explainxkcd.com/wiki/index.php/1028:_Communicati...

That PowerPoint made my irony meter explode!

Even with the notes, it doesn't make a lot of sense. Why would you illustrate communication problems with people who literally don't speak the same language?

It’s almost impossible to understand, just from looking at the comic. Some of his comics are like that.

I stared at it for quite a while, before it “clicked.”

Once it did, the message was obvious.

The PowerPoint was designed to be presented. It’s not particularly useful, just being read. It really needs a narrator that can explain the concepts.

The idea is that it’s possible to communicate effectively, even when it seems impossible, as long as we are willing to take responsibility for the message, and figure out how to make it work. In order to do that, we need to understand the message, the recipient, the context, and the medium.

It also shows how we can lose the message, when we get sidetracked by the messenger or the medium.

If you read the notes on each slide that becomes pretty clear.

> people who literally don't speak the same language?

That's not my understanding of the XKCD comic, if that's what you mean.

Also, I don't find the comic to be hard to understand. It takes a little work but it's pretty clear.

It’s kinda funny, but this is also covered in the presentation. Something may seem clear to you, but it’s not clear to others.

Ha. Great point.

It's a great presentation and deserves to be more widely disseminated, you should probably submit it to HN as a submission.

Thanks, but it would probably not be received too well.

It has been my experience, that folks really don’t like its message.

I recently wrote this to someone who tends to give commands instead of advice, which is along the same lines:

"You can lead a horse to water, and it may not drink. If you drag a horse to water, it will refuse to drink, then kick you, and run away."

Stealing it...

The reason is that most people wouldn't care about the actual thing happening. Mozilla chooses to frame things in a way that is more likely to motivate people to action, even if that means being vague or dishonest. In this case their point #2 is dishonest because it already works that way.

In the app there is a "Share" button at the top right. After clicking you see an interstitial with a big "Post" button at the bottom. When you click that button, the chat is shared.

The ends justify the means always backfires in communication. People just lose trust in what they are being told.

There are so many examples of this. There was a poster that attempted to reduce needle reuse by showing the same needle degrading after multiple uses. The problem is that was not very dramatic (also not where the risk lies) so they increased the magnification at each stage to make the degredation seem worse than it was. People recognised this and the primary message amongst their target demographic was anti-drug campaigners lie to you.

I'm not sure how much this will damage Mozilla. Perhaps not much because they have already lost so much mana. Before coming to the comments and reading this thread, I had already thought to myself, "Can I really trust what Mozilla says anymore?".

Perhaps that makes it even worse. To have doubts that are so quickly confirmed suggests not only that you can't trust them, but you reliably can't trust them.

I want to like and support Mozilla, I'm posting this from Firefox, that makes me one of the few sticking with them. They make it so hard sometimes.

The fact that the crafting of the message and its match with the audience is poor, doesn’t mean that “Ends justify the means approaches always backfire”.

You can make a whole media empire that just spends its time saying that “The other sides is lying” and taking on the role of pointing out every crack and flaw in a wall, lavishing hours upon hours of coverage.

You can find just the right audience to support you, who themselves believe that the ends justify the means. Willing to rethink their lines in the sand for their team.

Well yes, if your goal is to reduce trust in the things people are told. The side-effects amplify the effects.

The Mozilla organization is run poorly. I really wish it was different.

The Mozilla organization has leadership that is complicit with Google.

- Mozilla acts as an antitrust sponge for Google and in exchange gets lots of money

- Mozilla is encouraged not to make Firefox better than Chrome. They thus under-invest in Mozilla, Rust, and core browser tech.

- Mozilla spends all of its money on irrelevant efforts like VR, shitty platform plays, half-baked AI, and insane exec comp.

It's a drop in the bucket for Google's peace of mind.

If Mozilla wanted to be a next-gen serious company, they would become a developer tooling and platform company. They'd keep developing Firefox and Rust, they'd build up an ecosystem around Rust and WASM, and they'd be the best in the world at it. Deploy websites, micro services, and run CI with Mozilla.

Rust is infectious and growing, WASM will eat the web and the data center, and Mozilla is completely sleeping on it. What's sad is how much of a hand they had in developing it all.

They threw away the stuff that mattered.

>Mozilla is encouraged not to make Firefox better than Chrome. They thus under-invest in Mozilla, Rust, and core browser tech.

Is there any evidence of this kind of corruption? It always seemed like misguided altruism to me, or at worst a serious lack of focus.

I wish people had such faith in my own competence.

All it takes is one CEO. The ICs have nothing to do this.

> there any evidence of this kind of corruption

Follow the money. Google gives so many millions to Mozilla, it would be a miracle if that had no effect on their priorities.

It blows my mind how many people still publicly post venmo payments, so this doesn’t surprise me actually.

why venmo even has a "feed" is unknown to me. are people actually doom scrolling that?

If it's there people will use it like social media. "Ooh looks like Adrian went to La Bamba with Chilliwack, he paid them $50 for drinks there"

Engagement. It's part of a dark pattern that triggers that little neuron tie our tendency to compare ourselves to others, a sort of "keeping up with the Joneses" type thing. It is also billed as a free advertising for businesses (e.g. hey, look where your friends shop!) which encourages more businesses to accept Venmo as a payment method.

Anything for the sake of growth or perceived growth, up to and including privacy violations.

I figure whatever they are trying to achieve probably doesn’t work. Otherwise Cash App would do it. But important decision makers are in too deep to admit they were wrong. Smells of turning the Magic Mouse upside down to charge it.

CollegeHumor, now Dropout, made a skit about this very question: https://www.youtube.com/watch?v=BWFLztKBrLY

If i put your information in a feed, you'll look really stupid when you cry privacy violation down the road as you realize what I've been doing with your information.

It is definitely odd. I think it started off as a KYC-kind of check. If there’s some weird, possibly illegal reason you type into the “what is this payment for?” input, I read that someone on behalf of Venmo will contact you to have you explain it further and to investigate if it should lead to the closure of your account.

It is unsurprising that Venmo has a log of transactions, right? That’s a necessary part of the job. Having it as something that can be presented as a social feed is the weird thing…

It almost seems like a radical art project, philosophical statement, or social experiment around transparency. Like, hypothetically in some alternate universe if they did no KYC, and just published everybody’s transactions, your peers could inspect your transactions, the police could just look and see if you were transacting with criminals… sort of like open source transactions. Maybe that was the original idea? And then eventually they got some actual customers and said “shit we’re a real company now, let’s put the social experiment on the back burner, add an opt-out, and start doing in-house kyc.”

I had no idea this was true until a buddy of mine who I play hockey with started putting super offensive notes on the payment, trying to trigger someone since it was a hot debate after one of our games if this actually occurred. After about five or six of these, someone did in fact contact him first via email, then actually called him and asked him to explain the notes and yes they do monitor these and yes, if its really suspect, the feds will be notified.

Which then begs the obvious - if you're buying drugs, then don't put you're buying drugs or paying off your bookie.

I guess it's law enforcement on the honor system - when you do something illegal, you're expected tell the police-monitored feed that you did it. We assume that no one is so unethical that they keep their illegal acts secret.

For extra credit, let's put this stuff on the blockchain. Crime is solved!

Reply on news.ycombinator.com