The FTC win you haven't heard about
September 26, 2024 5:57 PM   Subscribe

AI smackdown: How a new FTC ruling just protected the free press by fighting back against AI-generated fake reviews. Every time someone gets caught posting phony AI-generated “best lists,” Uncle Sam is free to slap them with a bill for $51,744 per violation ... "with the rise in online reviews we have seen that bad actors can manipulate or fake reviews to deceive consumers for their own benefit." The ruling also bars product review suppression, compensation or incentives for creating customer reviews, and organizations that supposedly provide "independent" reviews - for their own products or services.
Why did the FTC have to go as far as making a whole new rule about fake reviews?

Because, whether they are on an Amazon product page or on a once-trustworthy media site, AI-generated fake reviews are among the most effective money-minting scams you can pull on readers — but to pull it off most profitably, the tech part of the equation requires you to use an already credible website as your puppet.

There are plenty of variations on this theme, but generally AI-generated reviews make money for a site in two ways. First, you deceive Google’s page-crawlers into thinking your reviews are providing a helpful and educational service of unbiased consumer journalism. That’s accomplished by gaming Google’s search engine optimization rules to get your AI-generated garbage to the top of a search results page for whatever product or service people are likely to be searching for.

Then you try to exploit the nuance of Google’s algorithm by gearing your site toward Google monetization under-the-hood — and then, finally, you repeat the process ad nauseam to generate a prolific flood of AI-generated copy.

Once you’ve done this, you are now effectively driving a flood of site traffic toward affiliate-revenue generating links.
posted by Greg_Ace (10 comments total) 19 users marked this as a favorite
 
There will be a suit in N.D.Tx., Connors will issue a nationwide injunction staying enforcement, will ultimately find that it violated the "major questions" doctrine, the Fifth Circuit will unanimously affirm, SCOTUS will deny cert.
posted by praemunire at 6:02 PM on September 26 [1 favorite]


praemunire: I’m puzzled why you would say that. It certainly falls within the FTC’s express mandate to prohibit deceptive trade practices, and so I don’t see the major questions doctrine in play here at all. I very much doubt that the Fifth Circuit would act in accordance with your prediction.
posted by PaulVario at 6:21 PM on September 26 [3 favorites]


I should add: the notion that this FTC action “protects the free press” is some extremely strange framing. I can see how this increases the value of a journalistic product by creating incentives for accuracy and punishing deception — but you could say the same about (for instance) increasing liability for defamation. Such changes are typically described as the opposite of protecting the free press, though.
posted by PaulVario at 6:27 PM on September 26


Now would Big Corp ever push monetization hard enough to engage in deceptive trade practices...?

They knew it was slimy, unethical, and wrong, and they're doing it anyway. They're going to claim precedent and are looking for loopholes even before the law is drafted, let alone before, or if, it goes into effect. I'm sure they're gearing up to start paying the lobbyists and probably a judge or two. Finally, $51K and change is hardly the cost of an advertisement. If there is an AI generated fake news (lie) that's lucrative enough, that amount of fine will simply be the cost of doing business.
posted by BlueHorse at 6:38 PM on September 26


As much as I was briefly thrilled to see this, how is it ever practically enforceable? What is the cost of identifying even a few of the most egregious examples and what team of expert textual analysis can catch more than a trivial few and how do they prove in a court of law that it is not just a language challenged east asian writing the review? The constitutionality is less the issue as how any actual case could be effectively prosecuted.

But good on the FTC for trying.
posted by sammyo at 7:31 PM on September 26 [2 favorites]


The idea that the "major questions" fiction has any rules other than "what do 5 or 6 christian mullahs supreme court justices not like" is very amusing. You'd have better luck arguing for consistency with bedtime stories about Orcs and Klingons and Fairies.

Law is whatever people with power want. Get used to that, its about to get more vigorous.
posted by No Climate - No Food, No Food - No Future. at 7:40 PM on September 26 [2 favorites]


I applaud the FTC, finally returning to its real mission. If enforced, this rule will really help.
posted by No Climate - No Food, No Food - No Future. at 7:41 PM on September 26 [3 favorites]


I’m puzzled why you would say that. It certainly falls within the FTC’s express mandate to prohibit deceptive trade practices, and so I don’t see the major questions doctrine in play here at all. I very much doubt that the Fifth Circuit would act in accordance with your prediction.

Talk to practically any major agency that's issued final serious regulations affecting corporations in the past four-ish years and get back to me. I recommend starting with the SEC.
posted by praemunire at 8:00 PM on September 26


What is the cost of identifying even a few of the most egregious examples and what team of expert textual analysis can catch more than a trivial few and how do they prove in a court of law that it is not just a language challenged east asian writing the review? The constitutionality is less the issue as how any actual case could be effectively prosecuted.

The FTC has subpoena power. That is much less the problem, though still work.
posted by praemunire at 8:02 PM on September 26


This is a definite win and absolutely should be celebrated. I have serious doubts about any ability to enforce it, but at least it's a start in the right direction.

Finally, $51K and change is hardly the cost of an advertisement. If there is an AI generated fake news (lie) that's lucrative enough, that amount of fine will simply be the cost of doing business.
The penalty is per infringement. As far as I can see, the culprits here are bombarding us with wave after wave of fake reviews, so that $51k will start to multiply fast. A single infringement would be nothing, but a couple of hundred gets you over the million mark. To the extent this is actually enforceable, large numbers of infringements are going to be the only thing worth the regulator chasing.
posted by dg at 10:53 PM on September 26


« Older Listening To The Sun   |   Your weekly dose of female fronted metalcore: Newer »


You are not currently logged in. Log in or create a new account to post comments.