Brave Creators is not how you enter the Brave index

Who reported it, and when we saw it

Key facts

• The Brave Search help section listed 29 pages in its navigation on 2026-08-09, including Brave Search Crawler, Search Operators and Brave Search Index & the Right to be Forgotten & to Object. None of them is a webmaster console, and https://search.brave.com/webmaster returns HTTP 404. Reproduce: curl -s https://search.brave.com/help | grep -o 'href="./help/[a-z-]*"' | sort -u
• The IndexNow participant file at https://www.indexnow.org/searchengines.json listed seven endpoints on 2026-08-09: bing, yandex, seznam, naver, yep, internetarchive, amazonbot. Brave is not among them.
• Brave's crawler page states the crawler "does not advertise a differentiated user agent" and that if a domain or page is not crawlable by Googlebot, Brave Search's bot will not crawl it either (checked on 2026-08-09).
• The Brave Creators registration article, stamped "January 27, 2025 19:48 Updated", names the verification folder .well_known with an underscore rather than the standard .well-known, requires the file to be called brave-rewards-verification.txt, and shows a token of the form brave-ledger-verification=<token>. The same article calls DNS verification "a good fallback if you have problems with file upload verification".
• The payout regions article shows "Last updated: August 18th, 2025" in its body and a help-desk stamp of "July 21, 2026 02:49". It lists 24 Uphold regions, none of which is an EU member state, and 29 regions where Solana self-custody is not supported, including Bulgaria, Croatia, Romania and Slovenia.
• Profound's post of 21 March 2025 reports 86.7 percent overlap - 13 of 15 results - between Claude's cited results and Brave's top non-sponsored results at p below 0.0001, across three commercial queries. Search Engine Land reported a later Profound figure on 17 June 2026: 79.2 percent of Claude's citations coming directly from Brave's top 10 results.
• Our own file at https://dablock.ai/.well-known/brave-rewards-verification.txt returned 146 bytes of prose with "Domain:" and "Token:" lines instead of the documented brave-ledger-verification= string on 2026-08-09, and dabyte.ai returned 404 at both the hyphen and the underscore spelling.

The two things that share a name

Brave Creators is a way to receive money from readers. The Brave Creator FAQ, stamped "February 13, 2025 20:36 Updated", describes two contribution types: an on-demand one-time contribution in BAT, USDC, ETH or Solana, and a recurring monthly contribution in BAT that repeats every 30 days. A custodial account is not required to receive contributions, but the blue checkmark shown to Brave users requires either a linked Solana or Ethereum self-custody address or a connected Uphold, Gemini or bitFlyer account. Registration was free when we set up a channel on 2026-08-06; Brave's documentation does not state a price either way, so treat the zero as our observation rather than a documented term. One thing to state precisely, because the loose version of it is wrong. Brave does not pay creators a share of its advertising revenue. Brave's rewards page says it shares ad revenue with users, in its words "sharing any ad revenue we generate on eligible ads directly with users", and separately lets those users send BAT contributions to sites and creators. So advertising money can reach you, but only because a reader chose to pass some of it on. The decision and the amount belong to the reader, not to Brave (checked on 2026-08-09). The Brave Search index is a different system with its own crawler, described on a single help page at search.brave.com/help/brave-search-crawler. Nothing in the Creators flow touches it. The verification token you place on your domain proves ownership for payout purposes; it is read by the Creators dashboard, not by the crawler. There is no field in the Creators dashboard that accepts a URL, a sitemap or a change notification.

Why people connect them anyway

The link people assume exists comes from somewhere real. Anthropic added Brave Search to its subprocessor list effective 19 March 2025, reported by Simon Willison on 21 March 2025 and by TechCrunch the same day; Willison also found a BraveSearchParams property in the web search tool definition. Profound published a measurement on 21 March 2025 finding 86.7 percent overlap between Claude's cited results and Brave's top non-sponsored organic results. Treat that number as directional rather than settled: it is 13 matches out of 15 results across three commercial queries, from one vendor, published the day after Claude's web search launched. Profound later re-ran the comparison. Search Engine Land reported on 17 June 2026 that 79.2 percent of Claude's citations came directly from Brave's top 10 results for the equivalent search, attributing the figure to Profound research presented by Josh Blyskal. That report does not state the sample size, we did not find a primary Profound post for the re-run, and we did not reproduce it, so treat the number as reported rather than measured. It is the more recent figure and it points the same way as the March 2025 one: where you sit in Brave's results is relevant to whether Claude sees you. From there the reasoning goes: Brave has a program for site owners, so joining it must be the way in. It is not. The one place where Brave itself puts Creators and AI in the same sentence is BAT Roadmap 4.0, published 9 July 2026, which announces a Creator Contribution Protocol intended to pay registered creators micro-royalties when participating AI products use qualified content. That is an announced payment scheme, not an indexing path, and it was an announcement rather than a shipped product when we checked on 2026-08-09. The same post says the Brave Ads Notification unit begins a gradual sunset by the end of 2026, which is a reason not to build anything long-lived on the current shape of the program.

What the index side actually gives a site owner

Four checks, all runnable in under a minute. First, https://search.brave.com/webmaster returns HTTP 404 - there is no console with coverage reports, crawl errors or query data. The help section itself is not thin: its navigation listed 29 pages on 2026-08-09, among them Brave Search Crawler, Search Operators, Rerank, Google fallback mixing and Brave Search Index & the Right to be Forgotten & to Object. What is missing is not documentation, it is a console. Second, https://search.brave.com/submit-url exists, but the crawler help page frames it narrowly: you submit a page for re-fetching after you have added a noindex directive, or you submit a URL that no longer exists so it can be delisted. It is also JavaScript-only; a plain fetch returns "This page needs JavaScript to function." Third, the IndexNow participant file at www.indexnow.org/searchengines.json listed seven endpoints on 2026-08-09 and Brave was not one, so the one-POST notification you may already be sending for Bing, Yandex, Seznam, Naver, Yep, the Internet Archive and Amazonbot does not reach Brave. Fourth, the user agent. Brave's crawler page states plainly that the crawler does not advertise a differentiated user agent, because sites that allow only Google would otherwise discriminate against it, and that a page Googlebot cannot crawl will not be crawled by Brave either. Third-party bot directories do list a Bravebot token with a full user-agent string, so the picture is not perfectly clean - but Brave's own documentation names no token, which means an allowlist keyed on a Brave user agent is built on something the vendor does not commit to. The rule that is documented is the Googlebot rule, so the operative check is your robots.txt entry for Googlebot. Finally, the Brave Search API at brave.com/search/api is a paid read interface - the Search plan was listed at 5 dollars per 1,000 requests on 2026-08-09 - and is not a submission surface.

Discovery is wider than inbound links

Our internal note said discovery into Brave happens only through inbound links. Checking it properly showed that is too narrow, so here is the correction. Brave's crawler page says the crawler is partially powered by the Web Discovery Project, an off-by-default opt-in setting in Brave browser's search settings. Brave's browser privacy page says WDP records some search terms and some visited pages that pass a series of filters, and that this data helps build the Brave Search independent index. MERJ published a reading of Brave's source at commit f25eb3d on 9 July 2026, updated 23 July 2026, that separates WDP into two channels: a page channel, where a page reported by roughly 20 opted-in users on separate networks becomes a discovery candidate, and a query channel, where a single opted-in user viewing a Google, Bing, Yahoo or DuckDuckGo results page on which your page ranks is enough. We have not reproduced their source reading, so attribute that structure to MERJ rather than to us. Their conclusions about the absence of a console, sitemap submission and IndexNow match what we measured independently. The practical consequence is worth stating plainly: your visibility in the mainstream engines is an input to Brave discovery, so the Brave work is mostly not Brave-specific work.

The verification traps we walked into

The Brave Creators documentation writes the verification folder as .well_known with an underscore, while the web standard and every other verification flow use .well-known with a hyphen. The article is stamped January 2025 and still reads that way on 2026-08-09. The safe procedure is to place the file exactly where the verification wizard points at the moment it hands you the file, then confirm with curl against both spellings before clicking verify. Our own site is the example of what happens when nobody runs that curl. On 2026-08-09, https://dablock.ai/.well-known/brave-rewards-verification.txt returned 146 bytes of human-readable prose with a "Domain:" line and a "Token:" line - not the brave-ledger-verification= string the documentation specifies - and dabyte.ai returned 404 at both spellings. A file was written by hand, in the shape someone assumed, at the path someone assumed, and never checked against the documented format. There is a second failure mode underneath it. Our releases replace the site tree atomically, so anything placed by hand under the web root disappears at the next deploy. That has cost us twice: an IndexNow key file, and /.well-known/mcp-registry-auth, which vanished silently and was noticed only when we needed to update a registry record and could not authenticate. For a site built by a generator, prefer the DNS TXT method - Brave presents it not as an equal alternative but as a fallback, in its words "a good fallback if you have problems with file upload verification", and a DNS record survives a rebuild. One more limit to plan around: the same article states that subdomains cannot be separately verified unless they are on the public suffix list, and that hello.example.com collapses into example.com. A network of properties on subdomains gets one channel, not many.

Payouts, and where the map ends

The payout side is where this stops being viable for a lot of readers. The regions article lists Uphold for 24 regions - Andorra, Argentina, Australia, Bahamas, Belize, Brazil, Cayman Islands, Chile, Colombia, Costa Rica, Grenada, Guatemala, Hong Kong, Monaco, New Zealand, Paraguay, Peru, Singapore, South Africa, Switzerland, Taiwan, United Kingdom, United States excluding New York state, and Uruguay. Not one of those is an EU member state. Gemini is United States only and Creators only; bitFlyer is Japan only; ZebPay is India only and appears for Rewards rather than Creators. Solana self-custody covers most regions but the article names 29 exceptions, and four of them - Bulgaria, Croatia, Romania and Slovenia - are EU members. Two further limits. Brave states you may connect a custodial account from an unsupported region, but you will only appear as a Verified Creator once your region is supported, so tell anyone you are advising before they start. And the broader list, "Unavailable Jurisdictions for Brave Rewards", sits behind authentication: on 2026-08-09 a GET to support.brave.app/hc/en-us/articles/9053832354957 ended at support.brave.app/auth/v3/signin with the article URL carried as the return_to parameter. That means a country question cannot be answered from public documentation - only by attempting registration. Size the channel honestly in whatever report you write. It pays what readers choose to send. With no Brave-using human audience, the expected amount is zero, and calling it AI monetisation is wrong on both halves: the payment is a reader's choice rather than a Brave payout, and nothing in it is triggered by an AI system using your content.

Checking your own coverage when there is no console

The site: operator is documented on search.brave.com/help/operators, with Brave's own example being "goggles site:brave.com". On 2026-08-09 that control query worked as documented: the results page showed "Only showing results from brave.com" and returned brave.com pages. The same day, site:dabyte.ai and site:dablock.ai both returned "Too few matches were found" together with a notice that search operators were not applied. Plain-keyword queries for the two domain names returned dabyte.com, a GitHub account, an unrelated Indonesian AI product and a Crunchbase profile for a differently-named company - not our pages. Read that result for exactly what it is. It is consistent with the two domains being absent from the index, and it is not a proof of a zero page count, because Brave dropped the operator rather than reporting a count. Two practical notes: repeated scripted queries get a captcha, so run this in a browser rather than in a loop, and since there is no console, the defensible line in any report is a dated screenshot of the query, not a coverage figure. We are the object lesson here. Two sites built specifically to be read by answer engines, with robots.txt allowing everything and a sitemap in place, are not findable by their own names in the index that Claude's web search draws on.

Questions this answers

Machine access

• /api/articles.json — every piece we published, with the measurement each one rests on
• /api/aiv.json — the AI Visibility Index the numbers above come from
• How the index is measured