Hacker News Reader

Comment by CSMastermind | Hacker News Reader

Recently I've been getting shown lots of deepfake ads on Facebook. I report them but then I get back a message saying that they don't violate the advertising guidelines. There's no option to report it specifically for being a deepfake.

I imagine this is a similar problem that Meta won't address until they're forced to.

View on HN

I found an entire company of 50 or so fake people called "AI FIRE" All the same bio. All generated images. Reported multiple profiles as fake LinkedIn as fake.

I only got halfway before it stopped letting me report.

Now LinkedIn doesn't let me report things anymore. It just gives me an "unknown error" message.

*AI FIRE, i got it backwards.

All the fake accounts remain. Because it's not in LinkedIns interest to reduce their usercount.

Do you have a link to the company? I'll see if I can report it as a LNKD alumni.

Sure, check the people images and bios https://www.linkedin.com/company/aifireco/people/

I am not 100% certain, but I think your report worked.

Someone looked into it, yes. Thanks for reporting this!

Is it a fake of a recognisable person? What are they endorsing/promoting?

I don't see the issue with using AI to generate video ads. Sure, it makes it unreasonably cheap to produce video. But the cost of displaying the ad is the main cost anyway.

The use of the term deepfake implies it was of a recognizable person. That's illegal.

And if that person has agreed, getting paid for the permission? It would be way more expensive to get $celebrity to have a shooting in a remote location compared to calling them/their agent for a permission. And the result could be comparable.

Yes: Joe Rogan, Donald Trump, and Mr. Beast are the ones I've reported so far.

One of them was a pill that supposedly grew your penis, another was for some conspiracy theory site, and one was just a straight up phishing/scam attempt.

Here's one of the products:

https://thepatriotsdigest.co/code-red-blackout

Here's one of the ads:

https://www.facebook.com/ads/library/?active_status=all&ad_t...

Apparently all of these were somehow reviewed by a human and determined to be within Facebooks guidelines.

Yeah don't these all supposedly go through a review process? How does that go through?

A team of facebook engineers could write an AI to at least flag that stuff for manual review in under a day.

The review process is just checking whether they're a paying customer

At one point I tested that theory by literally creating a paid Facebook ad for getcryptolocker.com, advertising that getting your data held to ransom was just one click away. I had no problems getting the ad published and Facebook alleges I had a reasonable amount of clicks.

> There's no option to report it specifically for being a deepfake

FB/Instagram is fueled by AI. It makes them money. They're not going to see this kind of 'pink slime' content as being a hindrance.

I commented a few weeks ago on something saying this was exactly what generated video was going to really take off in. Now you don't even need to pay the minimum wage for a struggling actor to endorse whatever thing you're trying to make: now you can just generate someone doing it. Or you know, a famous person.

Personally, going forward, this is going to be an automatic disqualification for me for any product or service. If you can't even be fucked to hire an actor, I'd imagine you can't be fucked to build whatever it is correctly either. No sale. I don't assume this will be a tectonic shift in my consumption habits anyway. I don't buy shit from instagram ads and the like and I have a feeling that's where this is going to really take off, but still.

On the other hand, maybe if they are efficient with their ads, perhaps they have more budget to put into the product instead of the marketing.

I would rather buy a product that put 90% of the budget into the product and 10% to ads as opposed to 50%/50%.

But I'm sure, they will first A/B test this with real actors to see what the results are before they decide on the course.

The problem is there's no way to differentiate on the consumer's end from well-prioritized internal spend while responsibly assuring quality where required, and just being a cheap ass business operator who's nickel and dimeing his own products into shit.

It feels like everywhere you look online is nothing but scams and grift now. Everyone out to fleece everyone else with minimum viable products. Nonsense letter combinations masquerading as brands all over Amazon hocking aliexpress garbage with poorly photoshopped product images comes to mind immediately.

Reply on news.ycombinator.com