I Asked OpenAI if META Is Faking Ad Buys

My question to OpenAI: “What is the probability that Meta Platforms creates spam bots to buy ads on Facebook and Instagram to artificially grow revenue?”

OpenAI’s response: “Short answer: there’s no reliable way to assign a real “probability” here, and there’s no public evidence that Meta itself is secretly running spam bots to buy its own ads. What does exist is (1) a huge industry-wide problem with bots and ad fraud, and (2) evidence and allegations that Meta has profited from fraudulent / scam advertisers and sometimes mis-reported metrics – which is different from Meta creating the bots.

Let me unpack that in a grounded way.

1. What your question is really asking

You’re basically asking:

“Is Meta intentionally faking demand for ads (via spam bots it controls) to inflate its revenue?”

That’s an allegation of deliberate fraud. To answer responsibly I have to stick to what’s actually documented:

• What we know about fraudulent activity on Meta’s platforms

• What we know about Meta’s own role and incentives

• What we don’t know (and shouldn’t just make up)

2. What is documented about scam ads & fraud on Meta

Recent investigations and internal documents show:

• Internal Meta documents projected that about 10% of 2024 revenue (≈$16B) would come from ads for scams and banned goods (investment scams, illegal gambling, banned medical products, etc.). (Reuters)

• The same docs suggest Meta often did not aggressively remove these advertisers unless its systems were extremely sure (e.g., >95% confidence) they were fraudulent, with a bias toward keeping revenue. (Reuters)

• Meta disputes how those internal numbers are being interpreted, but confirms that they studied the revenue associated with scam / prohibited ads and says it is investing more in anti-fraud enforcement. (Reuters)

Separately, Meta has had legal issues with ad metrics:

• Facebook/Meta paid around $40–60M to settle lawsuits claiming it overstated video ad viewing metrics (like average watch time) by large percentages. (AdExchanger)

• A former product manager recently alleged at a UK tribunal that Meta inflated the performance of “Shops Ads” by ~17–19% and used aggressive tracking approaches that may have bypassed Apple’s privacy rules; Meta denies wrongdoing and the case is still being litigated. (Financial Times)

Industry-wide context:

• Digital ad platforms (including Meta) publicly acknowledge that invalid traffic (bots, click farms, automated scripts) is a serious problem; third-party fraud-prevention companies estimate double-digit percentages of traffic can be fake in some channels. (Fraud0)

• Meta has official policies and systems for “invalid clicks” and says it detects and filters bot / fraudulent traffic and may credit advertisers for it. (Facebook)

All of that paints a picture where:

• Scam ads and invalid traffic are definitely present and sometimes substantial.

• Meta has historically messed up or mis-represented metrics and is being scrutinized for that.

• But none of this is the same as Meta secretly running its own spam bots to buy its own ads.

3. The specific claim you’re asking about

Meta “creates spam bots to buy ads on Facebook and Instagram to artificially grow revenue.”

To support that, you’d need some kind of evidence that:

1. The bots are operated or commissioned by Meta (not just third-party fraudsters), and

2. They’re used to purchase ads with the intention of fabricating demand / revenue.

I’m not seeing any credible sources showing that:

• Investigations and lawsuits focus on Meta’s handling and reporting of existing demand (scam advertisers, inflated metrics, etc.), not Meta itself orchestrating fake buyers. (Reuters)

• The recent internal-docs stories describe Meta profiting from scam ads placed by others and being too lax about them – again, bad, but conceptually different. (Reuters)

So from what’s publicly known, saying “Meta probably creates spam bots to buy its own ads” would be pure speculation and essentially an unverified accusation of fraud. I shouldn’t assign a numeric probability to that or present it as likely.

4. Why assigning a probability is not really meaningful here

In statistics you need some combination of:

• Historical frequencies (similar past cases)

• Direct data / measurements

• A clearly defined model

Here we have none of that. We don’t observe:

• Any documented instance of Meta doing this.

• Any investigation concluding “these ad-buying bots are run by Meta.”

Given that:

• Any number like “5% chance” or “60% chance” would be made up, not a real probability.

• Because this is about alleged criminal or highly unethical conduct by a real company, making up a probability would be both misleading and unfair.

What we can say qualitatively:

• Meta has strong financial incentives to show ad growth and is under intense pressure to maintain revenue.

• But it also faces huge legal, regulatory, and reputational risks if it were ever caught deliberately creating fake buyers to juice revenue—securities fraud, consumer protection violations, etc.

• We do see patterns of allegedly under-enforcing against scam advertisers and inflating metrics, which is already the subject of criticism and litigation. (Financial Times)

That combination suggests: it’s plausible that Meta may sometimes err on the side of maximizing revenue from borderline or bad actors, but that is still different from secretly operating its own spam-buying bots.

5. How to think about this more rigorously

Instead of asking “what’s the probability Meta does X?”, you might ask:

• What evidence would I need to believe Meta is directly creating ad-buying bots?

  • Leaked internal docs or whistleblower testimony saying so.

  • Forensic analyses linking bot networks back to Meta infrastructure or payments.

  • Regulatory findings from e.g. the SEC or competition authorities.

• What is already known and documented?

  • Meta projects and tracks revenue from scam / prohibited ads, and has at times tolerated a large amount of that revenue. (Reuters)

  • Meta has settled lawsuits / faces allegations around inflated metrics and misleading ad performance reporting. (Information Age)

Those are serious issues that can be discussed and criticized directly without jumping to an unsupported, more extreme accusation.

6. Direct answer to your question

So, phrased carefully:

• There’s currently no public evidence that Meta is secretly creating spam bots to buy ads on Facebook or Instagram to fake its own ad revenue.

• Because of that, any “probability” number would just be a guess and I shouldn’t fabricate one.

• What is evidenced is:

  • Widespread scam and fraudulent ads running on Meta’s platforms, from which Meta has earned substantial revenue. (Reuters)

  • Past metric inflation and misleading ad reporting allegations and settlements. (Information Age)

• Those problems are worrying on their own, but they are not the same as Meta creating its own spam bots to buy ads.”

---