Reaching ChatGPT and AI assistants with your press release. Presswire guide.

AI press release distribution: reaching ChatGPT in 2026

AI assistants find press releases the same way they find any other web page. A crawler fetches the page, a search index stores it and the assistant retrieves it when someone asks a question it answers. ChatGPT, Perplexity and Claude run their own search crawlers; Microsoft Copilot draws on Bing; Google’s AI Overviews, AI Mode and Gemini draw on Google Search. A release gets in if it sits on an open, indexed page and states facts worth quoting. Nobody can guarantee the quote.

That is AI press release distribution with the marketing taken out. The detail below comes from the AI companies’ own documentation, read on October 23, 2026 and listed at the end, and from Presswire’s own server logs.

How each AI assistant finds a press release

Each of the large AI companies publishes documentation for website owners. Read together, it describes two ways in: a search index the assistant queries, and a live visit when a user’s question sends the assistant to a particular page.

AssistantWhere it looksCrawlers the company documents
ChatGPTOpenAI’s own search featuresOAI-SearchBot for search; ChatGPT-User for pages fetched during a conversation; GPTBot for model training
PerplexityPerplexity’s search resultsPerplexityBot for search (Perplexity says it is not used for model training); Perplexity-User for pages fetched on a user’s request
Microsoft CopilotBing; Microsoft says Copilot and Bing search rely on the same crawling, indexing and ranking foundationBingbot
Google AI Overviews and AI ModeGoogle Search’s indexGooglebot
GeminiGoogle Search’s index, used to ground answers in Gemini AppsGooglebot; Google-Extended is a separate control for training and grounding
ClaudeClaude’s web search resultsClaude-SearchBot for search; Claude-User for pages fetched on a user’s request; ClaudeBot for model training

Three practical points follow.

Training and search are separate switches. OpenAI says GPTBot collects content that may be used to train its models, while OAI-SearchBot surfaces websites in ChatGPT’s search features, and that the two settings are independent. Sites that block OAI-SearchBot will not be shown in ChatGPT search answers, according to OpenAI, though they can still appear as navigational links. Anthropic draws the same line between ClaudeBot, which collects possible training data, and Claude-SearchBot, which it says improves the quality of Claude’s search results.

Some visits happen because a person asked. ChatGPT-User, Perplexity-User and Claude-User fetch a page during a conversation. OpenAI says robots.txt rules may not apply to those user-initiated visits, and Perplexity says Perplexity-User generally ignores them. Anthropic says its bots honor robots.txt.

Google and Bing have no separate AI index to get into. Google says there are no additional requirements for AI Overviews or AI Mode and no need for new AI text files or markup: a page has to be indexed and eligible to show a snippet in Google Search. Even then, Google adds, indexing and serving aren’t guaranteed. Bing’s webmaster guidelines say Bing and Copilot search experiences rely on the same core crawling, indexing and ranking foundation. Google also says AI Overviews and AI Mode may run several related searches across subtopics to build one answer, so a page that clearly answers one specific question has more than one way in.

One consequence matters more than the rest. A release published this week is not in any model’s training data. An assistant can only use it by finding it through search or fetching it live, so for news, AI visibility is a search question, not a training question.

What makes a press release more likely to be quoted

Being fetched is not the same as being used. The search companies publish a fair amount about the pages they select for AI answers, and most of what drives a press release’s AI visibility is what a good news editor would tell you anyway.

Put the news in the first paragraph

Bing’s guidelines tell publishers to put essential information near the top of the page and to avoid long introductions. A first paragraph that says who did what, where and when gives a retrieval system the answer without digging, and gives a journalist the same.

Make each fact stand on its own

Bing says pages are more likely to be selected for AI answers when facts are explicit and key statements don’t rely on implied content. In a May 2026 post on how its index supports AI answers, Microsoft described grounding as relying on “discrete, supportable facts with clear provenance.” In a release, that means the company’s full name rather than “we”, the product’s name rather than “it”, the year rather than “recently” and the figure rather than “significant growth”.

Date it and attribute it

An assistant answering a question about something current needs to know when it was said and by whom. Date the release, name the issuing organization in full and attribute every quotation to a named person, with their role. Google’s Article structured data documentation lists the headline, author and publication date among the details that help it understand a news page, and Microsoft’s grounding post warns that out-of-date facts produce wrong answers.

Write the numbers in the text

If your key figure lives only in a chart or an infographic, a system reading the text can miss it. Bing’s guidelines say images and video should reinforce the text, not be the only source of information needed to understand the topic, and recommend alt text and captions. Google’s guidance for its AI features makes the same point: important content should be available as text. Put the revenue, the percentage, the price and the date in a sentence.

Use the same names every time

Bing asks for clear and consistent names for people, organizations, products and places, and says clear naming improves citation accuracy. If the headline uses your company’s full name, keep it in the body and the boilerplate rather than switching to an abbreviation or a former name.

Keep to one announcement

Bing says pages focused on a single topic are more likely to be selected for AI answers. A release announcing a funding round, a new chief executive and a product update at once is harder to quote accurately than a release about one of them.

Edit it like news, not an advertisement

Bing’s guidelines ask for content that is original and authoritative, and warn that thin, ad-heavy or affiliate-only pages may lose ranking and AI-answer eligibility. Superlatives an assistant cannot check, such as “industry-leading” or “revolutionary”, give it nothing to repeat. A claim backed by a number, a named source and a date does.

Mark it up honestly

Structured data helps machines read a page, within limits. Google and Bing both say the markup must match the visible text. Google says it does not guarantee that features using structured data will appear, and Bing says markup may support clearer grounding but does not guarantee visibility.

Our guide on how to write a press release covers headlines, structure and quotes in more detail.

Technical checks before you publish

These sit with whoever runs the site where the release is published: your own newsroom page, your distributor or both.

  1. Allow the search crawlers. Make sure robots.txt lets in OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot and Bingbot. Whether to allow training crawlers such as GPTBot and ClaudeBot is a separate decision; OpenAI says its two settings are independent.
  2. Check the firewall and CDN, not only robots.txt. Bot-management settings can turn crawlers away even when robots.txt lets them in. Cloudflare’s documentation, for example, says that from September 15, 2026, its settings for blocking AI training also block crawlers that serve both training and search.
  3. Leave snippets on. Google’s nosnippet and max-snippet controls limit what Search can show from a page, and Google says a page must be eligible for a snippet to appear as a link in AI Overviews or AI Mode. Bing says NOARCHIVE keeps content out of Copilot answers and NOCACHE limits Copilot to the URL, title and snippet.
  4. Notify search engines when the release goes live. IndexNow lets a site tell Bing and other participating search engines the moment a URL is added, updated or deleted, and Bing’s guidelines recommend it. The IndexNow FAQ is clear that submitting a URL does not guarantee immediate indexing. Google is not among the participants listed on indexnow.org; it finds new pages through its own crawling and sitemaps.
  5. Keep the URL stable. Bing recommends stable URLs, with 301 redirects when a page has to move, and says stable URLs support long-term visibility.
  6. Measure it. Bing Webmaster Tools added an AI Performance report in public preview in February 2026; it shows how often your pages are cited in Microsoft Copilot and in AI-generated summaries in Bing. For other assistants, ask a fixed set of questions on a schedule and record the date and the answer, because answers change from one run to the next.

How Presswire handles AI visibility

Presswire has distributed press releases since 2001. The readers have changed: alongside journalists, AI assistants now look for current, trustworthy information about companies. Our part is to put trusted company information and updates where today’s AI systems look for them. This is what happens to every release we distribute:

  • Published on presswire.com and our Google News site. Every release appears on both.
  • Checked by editors. Every release is checked before distribution by Presswire’s editors, former BBC journalists.
  • Open to AI crawlers from publication. Release pages are open to AI assistants from the moment they are published. OpenAI’s OAI-SearchBot and ChatGPT-User, PerplexityBot, Anthropic’s ClaudeBot, Googlebot and Bingbot all fetch presswire.com pages successfully, as do the crawlers of Meta, Amazon and Apple, and nothing on the site blocks them.
  • Bing notified within seconds. Each release is submitted to Bing’s index through IndexNow within about a second of publication. Bing’s index is what Microsoft Copilot draws on.
  • Marked up as news. Each release page carries NewsArticle structured data stating the headline, the issuing organization and the publication date.

The logs show it working. In the week to October 23, 2026, ChatGPT’s live fetcher, ChatGPT-User (checked against OpenAI’s published IP ranges), read hundreds of presswire.com pages, including many release pages, and ChatGPT answers sent readers to release pages. According to Ahrefs, presswire.com pages had been cited close to 200 times in AI-generated answers as of October 19, 2026, including 44 times in ChatGPT, 35 in Perplexity, 31 in Microsoft Copilot, 10 in Google’s AI Overviews and 8 in Gemini.

It is not instant. In our logs, AI crawlers typically fetched a new release within hours to a day of publication; the fastest we have verified was about two hours. Fetching is not quoting, either. Whether an assistant uses a release depends on what people ask it and what else it finds. Nobody controls that, so we don’t promise it.

Other wires market AI visibility too. PR Newswire’s own page, read on October 23, 2026, advertises press releases “optimized for AEO and GEO”, JSON-LD markup and a report on brand mentions in AI answers. Whoever you use, ask the same three questions: can AI search crawlers read the release page, how quickly are search engines told it exists and what evidence is there that assistants use it?

Distribution costs $250 per release for one region (the UK, Europe, North America or Asia) and $400 worldwide. There are no word-count fees, and images and video are included. Releases are normally sent within one hour of receipt. Prices exclude VAT. Any VAT is shown at checkout before you pay. Every rate is on the Presswire pricing page, and how press release distribution works at Presswire explains the service in full.

AI visibility is a by-product of good distribution, not a replacement for it. The journalists who cover your field are still a release’s most important readers.

A checklist for your next release

  1. Lead with the news: who, what, where and when in the first paragraph.
  2. Use full names for the company, products, people and places, and keep them consistent.
  3. Date the release and attribute every quotation to a named person and their role.
  4. Write key numbers in sentences, not only in images, and give images alt text and captions.
  5. Keep to one announcement, and cut any claim you cannot support.
  6. Publish on a page that search crawlers can reach, with no snippet restrictions.
  7. Notify Bing through IndexNow, and keep the URL stable.
  8. Track results in Bing Webmaster Tools’ AI Performance report and with a dated set of test questions.

Sources

All read on October 23, 2026. OpenAI: Overview of OpenAI crawlers. Perplexity: Perplexity crawlers. Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler? Google: AI features and your website, Google’s common crawlers, including Google-Extended and Article structured data. Microsoft: Bing Webmaster Guidelines, Evolving role of the index: from ranking pages to supporting answers (May 6, 2026) and Introducing AI Performance in Bing Webmaster Tools (February 10, 2026). IndexNow and the IndexNow FAQ. Cloudflare: Block AI bots. PR Newswire: press releases optimized for AEO and GEO. AI citation counts: Ahrefs Site Explorer, figures for October 19, 2026. Crawler activity: Presswire server logs, October 2026.

Frequently asked questions

Can I get my press release into ChatGPT?

You can make it findable. Publish it on a page that OpenAI’s search crawler, OAI-SearchBot, is allowed to read, and write it with clear, dated, attributed facts. OpenAI says sites that block OAI-SearchBot are not shown in ChatGPT search answers. Whether ChatGPT then quotes the release depends on what people ask, and nobody can guarantee it.

Does blocking GPTBot keep my release out of ChatGPT?

No, according to OpenAI. GPTBot collects content that may be used to train OpenAI’s models; OAI-SearchBot is the crawler for ChatGPT’s search features, and OpenAI says the two settings are independent. You can block training and still allow search.

How long does it take an AI assistant to find a new press release?

There is no fixed time. On presswire.com, AI crawlers typically fetched a new release within hours to a day, and the fastest we verified was about two hours. Presswire notifies Bing through IndexNow within seconds of publication, but the IndexNow FAQ is clear that a submission does not guarantee immediate indexing.

Do I need an llms.txt file or special AI markup?

Not for Google. Google says you don’t need new machine-readable files, AI text files or special markup to appear in AI Overviews or AI Mode, and points instead to the same SEO best practices it recommends for Search, including structured data that matches the visible text.

How much does Presswire cost?

$250 per release for one region (the UK, Europe, North America or Asia) and $400 worldwide, with no word-count fees and images and video included. Every release is checked by former BBC journalists before distribution and published on presswire.com and our Google News site. Prices exclude VAT. Any VAT is shown at checkout before you pay.