Leaking YouTube creators' private videos • Hacker News Board

Re: Leaking YouTube creators' private videos

#401

Post by 27183 » Sun, Jul 05, 2026, 3:31 PM UTC

falsemyrmidon wrote: post #389

rfrey wrote: post #338

Earlier quoted context omitted.

If the toilets at all those firms started operating in reverse, very little would get built. The conclusion is that plumbers are extremely important, not that plumbing is de facto engineering.

Plumbers build all the pipes, but one of the aforementioned professional engineers created a drawing of where they would go and how the water and sewage would move through the building.

And crucially, independent regulatory bodies formed the building codes[0] which govern things like minimum acceptable pitch for waste lines, size and materials selection of pipes, acceptable pressures and temperatures, etc. Engineers can be held responsible for failing to design to these codes, and plumbers and other tradespersons can be held responsible for failing to build to them.

IMO what we really need in software is something like the various codes which govern fire safety, efficiency, etc of modern buildings. It exists in some safety critical domains but should be more widely applied.

The safety codes are where we accumulate the hard-won knowledge over time. Many of the rules in the codes were paid for with human lives lost.

[0] https://en.wikipedia.org/wiki/International_Code_Council

Re: Leaking YouTube creators' private videos

#402

Post by Aaron_NW » Sun, Jul 05, 2026, 3:35 PM UTC

Great point. As we use agents to write and summarize responses, we need to treat the source content as a prompt injection surface.

Curious if you have a specific method for checking the content beyond the overall instruction not to trust it?

Re: Leaking YouTube creators' private videos

#403

Post by AlotOfReading » Sun, Jul 05, 2026, 3:50 PM UTC

rTX5CMRXIfFG wrote: post #330

john01dav wrote: post #260

Earlier quoted context omitted.

> Honestly it’s hard to refute the fact that we need roads and houses more than we need cat videos. Software does more than cat videos. Examples that may be relevant: - CAD and simulation tools that physical world engineers use - telecommunications (not just programmers, but programmers are vital for the current ultra-cheap generation) - CT and MRI data processing - alphafold - scheduling systems for universities and…

Conveniently, those are the subdomains of the tech industry that require a deep understanding of CS theory (eg embedded systems) and DS&A, and yet whenever the subject of Leetcode comes up on HN, software “engineers” swoop in to argue how useless those are as a whole to 90% of the software being built in the world today… Which, really, implies that 90% of people in this industry work in bullshit jobs that don’t requi…

How does embedded require deep understanding of CS? I work on embedded systems (safety critical no less) and while I can point out how they're actually distributed systems or x theorem is applicable all day, my coworkers don't actually need that info to do their jobs.

C and the various C++ subsets called "C with templates" would be much less common than they are if "deep understandings of CS and DS&A" were required in day to day work.

Re: Leaking YouTube creators' private videos

#404

Post by 1313ed01 » Sun, Jul 05, 2026, 4:10 PM UTC

cess11 wrote: post #205

fathermarz wrote: post #157

Earlier quoted context omitted.

I think there is a fine line. YouTube is not critical software and no one’s life depends on the safety (putting mental health aside) of the code running. Some software engineers do however write code that is critical, but to your point, I don’t think they are ever considered liable. I went through an acquisition as a Canadian software developer getting acquired by an American company. They wanted us to be called engi…

Once I worked in a company that had an ex-Googler on the board, who insisted on calling us engineers and wanted us to call ourselves that. In swedish, of course, 'ingenjörer'. It's not a protected title in Sweden, but we still refused, because we were nothing like engineers. We were a minuscule team of mostly self-taught hackers who happened to be employed to solve business problems in a system for managing other com…

I worked for several American companies in Sweden an all of them insisted on calling us engineers. Many I work with happen to be engineers, but I dropped out before getting my (computer) engineering degree. It's embarrassing. I would never call myself a software engineer or any other kind of engineer (or a doctor or lawyer or anything else I am not).

Re: Leaking YouTube creators' private videos

#405

Post by lnrd » Sun, Jul 05, 2026, 4:17 PM UTC

Mg6yDfjp5U wrote: post #10

I recently left Google having worked on a number of projects with various YouTube teams. I think I can explain why it's being handled this way by YouTube. This is a fairly nuanced/involved issue, so the task of classifying the bug likely made it's way to one of the engineers responsible for the implementation of this feature. That engineer has already launched this project, and filed it away under their GRAD (perform…

In my company, user reported bugs get triaged and once they are confirmed they end up in the board of the team owning that part of the system (often they also built it, but is not necessarily the same people). Then there's a team bug threshold and if it gets reached then the whole team can't merge anything else until X amount of bugs are fixed and the number goes down. It's very annoying system and in cases of emergency it can of course temporarily be lifted by someone above, but honestly is very effective in making sure that bugs are fixed in a reasonable amount of time. This makes fixing bugs "part of maintenance work that is a given and expected to be done between new initiatives".

Re: Leaking YouTube creators' private videos

#406

Post by 27183 » Sun, Jul 05, 2026, 4:21 PM UTC

InsideOutSanta wrote: post #26

wrs wrote: post #5

>Comments should be passed to the model with clear role boundaries that prevent them from being interpreted as system-level directives. Well, such clear boundaries would solve lots of problems. But those don’t exist, do they?

Yeah, I suspect the main reason this was rejected is simply because it's not fixable. This is just how LLMs work. This LLM ingests untrusted data, so there will always be a non-zero chance that this type of prompt injection succeeds.

This makes me crazy. When I started my career in software the focus was on security, correctness, uptime (measured in nines--remember those?), and performance. Features are important, but building a crap feature is worse than building no feature at all.

I don't understand how these systems are passing the bar. You would have been fired for trying to railroad something like this into production 10-15 years ago. What happened?

Re: Leaking YouTube creators' private videos

#407

Post by 27183 » Sun, Jul 05, 2026, 4:32 PM UTC

thamzhack wrote: post #136

I've reported bugs to google VRP and got paid. The main problem with this report is that the victim has to click a suspicious link which is similar to phishing through email. No bounty programs award bounty for phishing. This is not to say this isn't a bug. The author has to find a way to escalate the impact. If they are able to achieve the same impact without user interaction the impact will be high enough for bount…

Isn't this more like an exploit which allows an attacker to send a phishing email from google's domain? They've hacked google's chatbot to send the attack vector. There's no way to justify it, google's behavior here is just crazy. User interaction isn't really the issue, it's that the attacker has appropriated google's brand to gain the user's trust. You'd think that might be something a company would care about? These are super weird times we're living in.

Re: Leaking YouTube creators' private videos

#408

Post by rurban » Sun, Jul 05, 2026, 4:46 PM UTC

NamTaf wrote: post #149

Mg6yDfjp5U wrote: post #10

I recently left Google having worked on a number of projects with various YouTube teams. I think I can explain why it's being handled this way by YouTube. This is a fairly nuanced/involved issue, so the task of classifying the bug likely made it's way to one of the engineers responsible for the implementation of this feature. That engineer has already launched this project, and filed it away under their GRAD (perform…

[flagged]

No, that's why Google does not do engineering. It's an ad company afterall.

Us engineers do have our priorities straight.

Re: Leaking YouTube creators' private videos

#409

Post by javxfps » Sun, Jul 05, 2026, 4:56 PM UTC

ireadmevs wrote: post #377

sothatsit wrote: post #274

Earlier quoted context omitted.

There is no data leak until a user clicks a suspicious link in the AI output. Clicking a suggested prompt alone does not have any risk of leaking data.

Yes, a link requires user interaction. But what if the attacker decides to render an image instead and put the secret data in the query params? Loading an image is a way to trigger a request without user interaction

I actually ended up doing that to see if they would change their mind, but they didn't really seem to care.

Re: Leaking YouTube creators' private videos

#410

Post by javxfps » Sun, Jul 05, 2026, 4:59 PM UTC

tyrust wrote: post #399

chii wrote: post #370

Earlier quoted context omitted.

The attack requires a third party to unknowingly click on the engineered URL that leak private video title. Not sure if it counts as a POC if you can only use your own channel to prove it works. But still, it would require a user interaction to click on the link to leak data - and google should acknowledge it as an issue, because an attacker should never be able to generate a link they control in a trusted/secure env…

I understand the purported vulnerability. What I am saying is that I would be surprised if the URL were rendered, let alone made it to the victim.

I ended up making it render automatically, but they didn't really seem to care.