[postexitus]

Tapping into AdTech is extremely hard, as it's hard driven by network effects. What you mean is "displaying ads inside OpenAI products" then, yes, achievable, but that's a miniscule part of targeted Ad markets - 2% is actually very optimistic. Otherwise, they can sell literally 0 products to existing players as they all have already established "AI" toolsets to help them for ad generation and targeting.

[friendzis]

Query: LibraGPT, create a plan for my trip to Italia

Response: Book a car at <totally not an ad> and it will be waiting for you at arrival terminal, drive to Napoli and stay at <totally not an ad> with an amazing view. There's an amazing <totally not an ad> place that serves grandma's favorite carbonara! Do you want me to make the bookings with a totally not fake 20% discount?

[xiphias2]

I'm traveling like this all the time already, I don't understand why it's hard for people to understand that ad placement is actually easier for chat than search

[KaiserPro]

> that ad placement is actually easier for chat than search

Yes, but the reason why people are turning to chatgpt is because the time to actual info that _I want_ is much much lower.

The point of advertising is to displace the thing that you actually want with something they are paying the company to promote.

You can handwave about personalization, but do you want adtech people having access to your life's context?

[tantalor]

> all the time already

What are you actually saying? You're already using chatbots that are embedding non-disclosed paid endorsements? And you like that?

> ad placement is actually easier for chat

Can you point to, I don't know, anything to back this up?

[poloniculmov]

But who wants that? And you're going to say that's exactly what a travel agent does, selling me stuff so he can get a kickback. But when stuff goes wrong, I'll yell at the travel agent so he has some incentive to curate his ads.

[tantalor]

How to speedrun massive penalties and disgorgement from FTC.

I guess we'll just put that in the "Cost of Goods Sold" bucket.

[mh-]

I'm not aware of any FTC rule that would preempt this sort of product as long as it met the endorsement disclosure rules (16 CFR Part 255), same as paid influencers do today.

What are you imagining they run afoul of?

https://www.ftc.gov/business-guidance/resources/ftcs-endorse...

https://www.ecfr.gov/current/title-16/chapter-I/subchapter-B...

[tantalor]

friendzis's example showed a plausible way to generate revenue by inserting paid placements into the chat bot response without disclosures by pretending they are just honest, organic suggestions.

[mh-]

Right. That's not a novel idea, and this is a well-trod area of concern. That's why these FTC rules have been around for many years.

edit: to be clear, I am saying that in the absence of clear disclosures, that would run afoul of current FTC rules. And historically they have been quick to react to novel ways of misleading consumers.

[friendzis]

All these chatbots are openly making recommendations for particular products since the day one. FTC (or any other regulatory body) does not even look at that direction.

Do you have at least a rough idea how many current product recommendations are influenced grok "musk is the bestest at everything" style?

[postexitus]

Let's put an analogy to Google ads - the ads that appear at search results do not make up even 5% of their ad revenue. Even smaller for Meta. They earn their big ad revenues from their network, not from their main apps.

[hobofan]

What? Where are you getting those numbers from?

Every source I know (hard to link on mobile) shows Google Search to make up 50+% of their ad revenue, and there has been extensive reporting over the years on Google's struggle to diversify away from that.

[PunchyHamster]

bribe the artificial idiot to con the user, brilliant !

[lesuorac]

That seems a bit risky for when the car isn't waiting for you at the terminal.

At least with an ad it's obvious a separate company is involved. If you do all the payment through OpenAI it seems to leave them open to liability.

[morkalork]

Could be as simple as referral link commission like all those totally-not-content-farm travel blogs.

Travel sites, VPNs and insurance all pay quite handsomely (compared to say amazon links on cooking sites)

[friendzis]

> If you do all the payment through OpenAI it seems to leave them open to liability.

Booking, airbnb, rentalcars, etc all seem to be doing pretty fine regulatory wise.

[echelon]

If ChatGPT shows ads, I'll switch to Claude or Gemini or DeepSeek.

[tomashubelbauer]

I expect all hosted model providers will serve ads (regular, marked advertisements, no need for them to pretend not to, people don't care) once the first provider takes the lid off on the whole thing and it proves to be making money. There's no point in differentiating as the one hosted model with no ads because it only attracts a few people, the same way existing Google search and YouTube alternatives that respect the user are niche. Only offline, self hosted models will be ad free (in my estimation).

[HWR_14]

Assuming you know it's an ad. Ads in answers will generate a ton of revenue and you'll never know if that Hilton really is the best hotel or if they just paid the most.

[mh-]

This isn't a realistic concern unless FTC rules changed substantially from where they are today (see my other comment on this post for links). Sponsored link disclosures would be in place.

Everything else aside, it's simply not worth it for them to try to skirt these rules because the majority of their users (or Google's) simply don't care if something is paid placement or not, provided it meets their needs.

[HWR_14]

That's only true if you can demonstrate a substantial percentage of people would be unaware of it. The reason influencers have to disclose is some but not all take endorsement money. It would be pretty easy for OpenAI to say it was common knowledge or to bury disclosures in the fine print of the services and not every time it happened

[sjsdaiuasgdia]

The US federal government is now a mob-style organization. The laws, rules, and regulations that are written down are only applicable as far as Trump and those around him want them to be. Loyalty to the boss is the only inviolable rule.

In other words, if they want to put ads into chat, they just need to be perceived as well aligned to Trump to avoid any actual punishment.

[wood_spirit]

The OpenAI pitch for “publishing partnerships” (basically buying bias and placement) leaked last year.

[dooglius]

They're all already trained on ads, and it would be silly to think advertisers aren't going to optimize for this.

[lambda]

Do you think it would show ads, or just prioritize content based on who has paid for ad placement?