Competitors Are Bidding on Your Prompts: How to Detect AI Search Ads Targeting Your Brand in 2026

Key takeaways

• ChatGPT ads went live in the US on February 9, 2026 and opened to self-serve on May 5. Competitors target conversations about your brand using "context hints," and OpenAI's ad policies have no trademark complaint process to stop them.
• Native reporting shows daily impressions, clicks, and spend, nothing else. With no equivalent of Google's Auction Insights, detection has to happen from outside the platform.
• Early testing by Seer Interactive found branded prompts largely uncontested. A rival can claim the sponsored slot on your own brand terms without outbidding anyone, because you haven't bid either.
• Most AI visibility tools track organic citations only. Paid placements at prompt level are a separate dataset that only a few tools cover, so check before you buy.
• Detection is half the job. Claiming your branded prompts, holding the organic answer, and tracking share of voice weekly is what protects revenue.

The ad channel nobody can audit

OpenAI started serving ads inside ChatGPT answers on February 9, 2026, for logged-in adults on the Free and Go tiers. What began as a pilot with a $200,000 minimum commitment became a self-serve Ads Manager on May 5, open to US businesses, with CPC bidding added alongside CPM. Pilots in Canada, Australia, and New Zealand followed in late March. The UK, Mexico, Brazil, Japan, and South Korea went live by August 11.

Advertisers noticed. Adthena counted more than 600 of them by May, including Best Buy, Target, AT&T, and Expedia, against a user base of roughly 800 million weekly actives. Across nearly 1 million query indexes Adthena analyzed between March and May, ads appeared on 4.5% of US queries. Around 90% of all placements in that dataset were US, Australia sat at 1.6%, and the UK was still at zero, waiting for its switch to flip.

Here's the part that should bother you. In Google Ads, conquesting has always been visible. Auction Insights tells you who overlaps with you, and the search terms report shows when your ad fires on a competitor's brand. In ChatGPT, an advertiser gets daily impressions, clicks, spend, CTR, and average CPC or CPM. That's the whole report. You cannot see which prompts triggered anything, who else appeared, or how your presence compares to the market.

Your competitors are running ads on prompts that mention your brand. You can't see them. They can't see yours either, which is small comfort.

The timing matters beyond ChatGPT. Google's AI Max campaigns replaced keyword-level control with semantic intent matching, Google already runs ads inside AI Overviews, and Microsoft Advertising is positioning its ad stack for the same shift. The 2POINT agency's 2026 guide projects AI-driven search advertising growing from about $1.1 billion in 2025 to $26 billion by 2029. Whatever the exact number, the direction is clear, and the detection problem will follow the money into every engine that sells ads.

How competitors target your brand in AI answers

Context hints replaced keywords

There is no keyword auction in ChatGPT ads. Advertisers write "context hints," which are descriptions of the kinds of conversations where their product is relevant, then supply ad copy and a landing page. OpenAI's system matches against the live conversation, and ranking blends relevance with bid. There's no quality score feedback, no hint-match reporting, and no way to see which conversations triggered your ad.

The practical effect: a competitor doesn't bid on your brand name the way they would in Google. They describe a context, something like "conversations where someone is comparing tools for X" or "people asking about alternatives to Y," and the matcher decides when their ad shows. From the outside, the result looks identical to conquesting. Their ad appears under prompts that mention you.

The conquesting loophole

OpenAI's ad policies, updated August 31, 2026, prohibit counterfeit goods and misleading comparative claims. What they don't prohibit is bidding into competitor brand contexts. There is no equivalent of Google's trademark complaint process, so if a rival wants to appear under "[your brand] alternatives," policy will not stop them.

One design choice softens the blow. OpenAI calls it Answer Independence: ads render below the response, clearly labeled Sponsored, and cannot influence the answer itself. On a branded prompt, the model can recommend you while a competitor's ad sits directly beneath that recommendation, betting you'll lose the click anyway. There's something almost funny about the arrangement. The machine vouches for you, and someone else's ad waits right under it.

Adthena's placement index, built on roughly 50,000 daily ad placements, shows what that ad looks like: a headline around 30 characters following a "Brand: Benefit" formula, a body of about 19 words, hard numbers and dollar figures outperforming vague claims, and "Free" as the most common hook. One brand can even appear twice in a single response, a quirk Adthena calls being "double parked."

What the ad platform tells you, and what it hides

The Google comparison is worth dwelling on because conquesting detection in paid search is a solved problem. The search terms report flags when your ads fire on competitor brand queries, which Ryze's analysis notes typically convert at 2 to 5% of your branded traffic rate. Tools read that report for you automatically. None of that infrastructure exists on the AI side yet.

How to check for competitor ads manually

You can get a rough read without any tool. The catch is that most people test wrong and conclude they're safe when they aren't.

• Use the right account. Ads only serve to logged-in adults on the Free and Go tiers. Plus, Pro, Business, Enterprise, and Edu accounts never see ads, and ads never appear in Temporary Chats or the Atlas browser. If you checked once from your paid account, you saw nothing, and that told you nothing.
• Test from the right place. The US leads, Canada, Australia, and New Zealand have been active since March, and the UK, Mexico, Brazil, Japan, and South Korea joined by August. In the EEA, ad personalization is off by default, which changes what serves.
• Run prompts repeatedly. With ads appearing on only 4.5% of US queries in Adthena's sample, one clean result proves very little. Check the same prompts across several days and sessions before drawing conclusions.
• Know what you're looking for. A "Sponsored" block below the answer, a competitor or aggregator inside it, or the same brand twice.

Which prompts to run:

Manual checks are a sanity check, not a system. They can't cover thousands of prompts, and they can't compute share of voice. That's what tools are for.

Tools that detect paid placements in AI answers

Here's a distinction most buyers miss. The AI visibility market is crowded with tools that track organic mentions and citations, meaning whether models recommend you and which pages they cite. Paid placements are a different dataset. A tool that only watches organic answers will never see a competitor's ad, because ads don't touch the answer at all.

A few platforms track both sides.

Promptwatch is one of them. Its Ads Radar shows which competitors bid on your prompts, what ad copy they run, and how they position themselves, and its ChatGPT Shopping tracking catches product recommendations where a rival's offering appears instead of yours. Because it reads the actual user interfaces of ChatGPT and the other major engines rather than API outputs, it sees sponsored placements the way users actually see them.

Adthena's ChatGPT Ads Intelligence was the first whole-market tracker and remains the deepest on the paid side. It monitors 300,000+ prompts daily, reports prompt-level ad detection rates, competitor counts, and share of voice week over week, and includes a greenfield filter for high-intent prompts where nobody is advertising yet. It costs $399 per month flat with a 21-day trial, and it's built for teams already spending seriously on paid search. The catch: it's paid-only, so you'll want organic coverage from somewhere else.

Otterly.AI covers ads and shopping across seven engines, including Google AI Overviews ad tracking, from $29 per month. The tradeoff is depth per engine.

Semrush sells its AI Visibility Toolkit as an add-on at $99 per month for one domain and 25 prompts, with competitor benchmarking. Useful if you already live in Semrush, lighter on paid placements.

Profound, at enterprise custom pricing, is strong on the organic side, with answer share and prompt volumes at scale. Pair it with a paid-specific tracker if conquesting is a live concern.

What to do when a competitor is on your prompt

Confirm it's conquesting, not category competition

A rival appearing under "best project management software" is competing for the category, same as always. Conquesting shows a pattern: their ad appears under "[your brand] alternatives," "[your brand] vs [competitor]," "[your brand] pricing," and your pure brand prompts. Judge the pattern across a prompt set, never a single sighting, because with context hints one appearance can be sloppy matching.

Claim your branded prompts while they're cheap

Seer Interactive's early testing found branded prompts largely unclaimed. No one is losing a bidding war here. The slot sits uncontested, and a rival can take it without outbidding anyone. So can you, which is the point. Claiming your own brand contexts now is the cheapest brand defense you will ever buy on this channel. Reported CPMs run from about $20 to $80 with CPCs of $1.50 to $6.00, and those numbers will not hold as the UK and other markets switch on and the advertiser count climbs past its current 600.

Own the answer, not just the ad slot

Answer Independence cuts both ways. A conquesting ad can't change what the model says, so the strongest position is holding the organic recommendation and the sponsored slot at the same time. The organic half is GEO work: making sure the model cites you, recommends you, and links to your pages when users ask about the category. The citation trends and content gap features in tools like Promptwatch exist for exactly this, and one agency reports brands combining AEO with paid ChatGPT ads seeing roughly double the conversion rates of either alone. Treat that last figure as a vendor claim and test it yourself, but the logic is sound.

Aggregators deserve their own line in your notes. At 11.6% of observed ads, they're often the ones taking category prompts, and they win by compiling other people's opinions. The counter is showing up in the sources they compile from, which in 2026 means Reddit threads and YouTube videos as much as your own domain.

Track share of voice weekly

One screenshot of a competitor ad is an anecdote. Share of voice week over week is a trend, and trends are what justify budget. Watch which advertisers appear on your prompt set, how often, and which high-intent prompts still have no ads at all. Those greenfield prompts are where you can buy attention before anyone else notices it's for sale.

A weekly detection routine

Keep it small enough to actually happen:

• Monday: run 10 to 15 core prompts from a Free-tier account in your main market. Screenshot anything marked Sponsored. Ten minutes.
• Weekly: review your monitoring tool for new advertisers, new prompts triggering ads, share of voice shifts, and ad copy changes. Most platforms, Promptwatch included, can push a weekly digest instead of making you pull.
• Monthly: review greenfield prompts, check aggregator presence on category terms, and feed what you found into both the paid plan and the content plan.

If you want AI search signals alongside competitor moves from every other channel in one feed, Competitors App does that kind of cross-channel logging.

Caveats worth knowing

Early benchmarks conflict, and that itself is information. Live testing by Jolly Good Web put CTR near 1% regardless of creative, while agency 2POINT claims 1.5 to 3.5%. CPM figures range from $25 to $60 per one agency and $20 to $80 per another, with Jellyfish reporting closer to $30 in practice. Treat every number from this channel as directional for now.

Ads only reach Free and Go tier users, a younger, more price-sensitive crowd. If your buyers live on Plus or Team plans, conquesting risk is lower today. So is the value of defending, and so is the reach of your own ads. B2B brands should factor that in before panicking or spending.

You also can't opt out of anything. No negative keywords, no placement exclusions, no category controls beyond OpenAI's defaults. If a competitor's ad appears near your brand contexts, there is no form to file.

And the channel is still consumer-first: lifestyle and household, local services, travel, digital products, and education, with finance, healthcare, and legal approved case by case. If you're outside those verticals, monitor, but don't expect much action yet.

The bottom line

The auction is months old, the reporting is a daily totals table, and the platform will never show you who is bidding on conversations about your brand. Detection is an outside job, done with a free account and a monitoring tool, checked weekly.

Run your brand prompts from a free ChatGPT account this week. If a competitor's ad is already there, you're late, and at least you know it now. If the slot is empty, claim it before someone else does, because right now nobody is making them outbid you.