What Did Reach Say About OpenAI?
In an interview published by Digiday on October 16, 2026, Reach chief executive Piers North described the publisher’s Google AI deal as a template and named OpenAI as the company it has not reached terms with. “We don’t have that same relationship with OpenAI. They’re one of the parties we need to bring to the table. We have sent them the next stage in the legal process,” he said. Digiday’s introduction calls it “a cease-and-desist letter” (Digiday).
North said Reach has “pretty good evidence – via our access logs – of how much they [OpenAI] are using our content,” that “we put up plenty of ‘do not walk on the grass’ signs,” and that “there’s a limit to how much technically you can do because the scrapers are getting much more sophisticated.” Of OpenAI, he added: “we can see how millions of times a month your bots are literally all over our content.” Digiday reports, “per the publisher,” that Reach “has seen well over 100 million successful scrapes of its content across its portfolio over the last six months.”
Fairness requires three notes. First, these are Reach’s claims about its own logs; we have not seen the logs or the letter. Second, the 100 million figure is not attributed to OpenAI; Digiday presents it as scraping across Reach’s portfolio. Third, Digiday’s article includes no response from OpenAI, and we found no public OpenAI statement about Reach. A legal letter is a demand, not a finding.
What Do Reach’s Sites Tell OpenAI’s Crawlers?
The “do not walk on the grass” signs are public. Every website can publish a robots.txt file telling named crawlers which paths they may fetch. We read the files for nine Reach titles on October 16, 2026: the Mirror, Daily Express, Daily Star, OK!, Irish Mirror, Manchester Evening News, Liverpool Echo, Birmingham Mail and WalesOnline. Each contains the same two groups:
User-agent: GPTBot / Disallow: / and User-agent: OAI-SearchBot / Disallow: /
number of the nine titles that fully disallow the token
Live read; robots.txt files can change at any time. Express is the one title that does not name Meta-ExternalAgent. A zero means the token is not named, so the general rules for all crawlers apply to it. All nine also block CCBot and several other AI tokens; none blocks Googlebot or Bingbot.
The chart shows that the block is not aimed at OpenAI alone: Anthropic’s, Perplexity’s, Apple’s training token and, on eight of the nine titles, Meta’s crawler are refused too, while Google-Extended and OpenAI’s user-triggered ChatGPT-User are not named anywhere.
The OpenAI block is not new. In the Internet Archive’s copies of the Mirror’s robots.txt, neither OpenAI token appears on January 1, 2024, and both are disallowed by October 2, 2024 (Wayback Machine). Reach’s current dispute is therefore not about whether it has opted out under OpenAI’s system. It has. It is about whether that opt-out is the end of the matter, and North’s point that robots.txt is a request, not a lock: it works only for crawlers that read and honor it.
What Does OpenAI Say Its Crawlers Do?
OpenAI publishes a page on its crawlers and user agents. It says it uses “OAI-SearchBot and GPTBot robots.txt tags to enable webmasters to manage how their sites and content work with AI,” and that “each setting is independent of the others” (OpenAI). The three agents it lists do different jobs:
| Agent | What OpenAI says it is for | What OpenAI says robots.txt does | Reach titles |
|---|---|---|---|
| OAI-SearchBot | “used to surface websites in search results in ChatGPT’s search features” | “Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links.” | Blocked on 9 of 9 |
| GPTBot | “used to crawl content that may be used in training our generative AI foundation models” | “Disallowing GPTBot indicates a site’s content should not be used in training generative AI foundation models.” | Blocked on 9 of 9 |
| ChatGPT-User | visits a page when a user asks ChatGPT or a custom GPT a question; “not used for crawling the web in an automatic fashion” | “Because these actions are initiated by a user, robots.txt rules may not apply.” It “is not used to determine whether content may appear in Search.” | Not named on any |
OpenAI adds that “it can take ~24 hours from a site’s robots.txt update for our systems to adjust” for search results. The page does not address content obtained through third parties, such as syndication partners or copies on other sites, and we do not speculate about what Reach’s logs show. Each side’s position is consistent with its own documents: OpenAI publishes opt-out controls and says they are honored; Reach says it has used them and wants a licence regardless.
Will Coverage in a Reach Title Show Up in ChatGPT?
Taken at OpenAI’s word, not as a cited search answer. A company featured in the Mirror, the Express or a Reach regional should expect that article not to be shown in ChatGPT’s search answers, because the site has opted out of OAI-SearchBot. OpenAI’s documentation allows one exception, “navigational links,” which it does not define further. We did not test ChatGPT’s answers for Reach articles, so this is what the rules predict, not what we observed.
Two other routes remain open in principle. A user can paste a Reach link into ChatGPT, and the ChatGPT-User agent is not one Reach names in robots.txt; OpenAI says robots.txt “may not apply” to such user-initiated visits. And the same story may be reported elsewhere: a quote that appears in the Mirror and also in a trade title or a wire story can reach ChatGPT through the outlet that permits it. Which sources AI assistants tend to cite is covered in Which Sources Do AI Assistants Actually Cite? and, for news sites more broadly, Do News Sites Block AI Crawlers? We Checked Fifteen.
Google is a different case. Reach’s files do not name Google-Extended, and Google says that token governs Gemini training and grounding, not Search; AI Overviews and AI Mode follow Googlebot, which Reach allows. The token-by-token detail is in How to Block or Allow AI Crawlers in robots.txt. Reach’s paid deal with Google’s News AI pilot, which Google says now has more than 400 publications, is in yesterday’s guide.
Where Does This Sit Among Publisher AI Disputes?
It is the latest of several publisher moves against OpenAI in different forms: some publishers have licensed content, some have sued, and Reach has, by its account, written. USA Today Co.’s lawsuit is covered in Why Did USA Today Co. Sue OpenAI?, and the licensing side in Which Publishers Have AI Licensing Deals? North framed the choice in commercial terms: “We can hold out for an optimum time or invest time and money in legal and other routes, or we can get on in shaping the market and learning together with partnerships like Google.” He said Reach is in discussions with Amazon, Google, Meta and Microsoft. One detail in the interview is out of date: it says Google has more than 200 publications in its News AI pilot, a figure Google raised to more than 400 on October 15.
What This Means for Anyone Buying a Placement
Our recommendations, not platform rules. First, when AI visibility is part of the brief, check the target outlet’s robots.txt before the pitch, not after the piece runs. It takes a minute: open the domain followed by /robots.txt and look for OAI-SearchBot, PerplexityBot and the other tokens in the chart. Second, do not let that check decide everything. A Mirror or Manchester Evening News piece still reaches those outlets’ readers and still appears in Google, which Reach permits; the question is only which AI products can cite it. Third, if ChatGPT visibility matters, pair a placement on a blocking outlet with coverage, or your own pages, that OAI-SearchBot can read. Fourth, be wary of anyone who promises a ChatGPT citation from any placement; on OpenAI’s own rules, for this group of sites, that promise cannot be kept.
Methodology and Limitations
Four limits apply. First, the robots.txt results are a live read of nine Reach domains on October 16, 2026, not every Reach site, and the files can change. Second, Reach’s legal letter, its log evidence and the 100 million figure come from one interview and are Reach’s account; OpenAI’s side is represented by its published documentation, since we found no statement about Reach. Third, we did not test ChatGPT’s search answers. Fourth, the Wayback Machine dates bracket when the Mirror added its block; they do not pinpoint it.
Conclusion
Reach has said no to OpenAI twice: once in robots.txt, by October 2024 at the latest, and now, it says, in a legal letter. For a company placing stories, the practical consequence came with the first. On OpenAI’s published rules, coverage in Reach’s titles should not be expected in ChatGPT search answers, however good the piece. That is worth knowing before a placement is sold or bought partly on the promise of AI visibility.
Frequently Asked Questions
Has Reach sent OpenAI a cease-and-desist letter?
Reach CEO Piers North told Digiday on October 16, 2026: "We have sent them the next stage in the legal process." Digiday describes it as a cease-and-desist letter. The letter has not been published, and the article includes no OpenAI response.
Does the Mirror block ChatGPT?
The Mirror's robots.txt, read on October 16, 2026, disallows both of OpenAI's crawlers, GPTBot and OAI-SearchBot, across the whole site. The same blocks appear on the eight other Reach titles we checked.
Can a Mirror or Express article appear in ChatGPT search answers?
OpenAI says sites opted out of OAI-SearchBot "will not be shown in ChatGPT search answers, though can still appear as navigational links." Reach's titles have opted out, so a cited answer from them should not be expected.
What is the difference between GPTBot and OAI-SearchBot?
OpenAI says GPTBot crawls content that may be used to train its foundation models, while OAI-SearchBot surfaces sites in ChatGPT's search features. Each can be allowed or blocked separately in robots.txt.
Does robots.txt stop ChatGPT from reading a page a user shares?
Not necessarily. OpenAI says that for ChatGPT-User, which acts on a user's request, "robots.txt rules may not apply." None of the nine Reach titles we checked names ChatGPT-User.
Does blocking OpenAI affect a Reach article in Google?
No Reach title we checked blocks Googlebot or Google-Extended. Google says Google-Extended does not affect Search, and AI Overviews and AI Mode follow Googlebot.
Sources
This explainer is built from Digiday's October 16, 2026 interview with Reach's chief executive, OpenAI's crawler documentation, a live read of nine Reach titles' robots.txt files and two Wayback Machine copies of the Mirror's file, all read on October 16, 2026. The robots.txt counts are our own parsing of the published files. Recommendations are ours. We did not test ChatGPT's answers, and we have not seen Reach's letter or logs.
- trade press Digiday, Reach CEO touts Google AI deal as a template, steps up pressure on OpenAI over scraping. Supports North's quotes, the legal-letter description, the 100 million figure and the absence of an OpenAI response. Dated October 16, 2026. Read October 16, 2026.
- platform documentation OpenAI, Overview of OpenAI Crawlers. Supports what GPTBot, OAI-SearchBot and ChatGPT-User do and how robots.txt applies to each. Read October 16, 2026.
- live read Reach / Mirror, mirror.co.uk robots.txt. Supports the Mirror's blocks; the same check was run on express.co.uk, dailystar.co.uk, ok.co.uk, irishmirror.ie, manchestereveningnews.co.uk, liverpoolecho.co.uk, birminghammail.co.uk and walesonline.co.uk. Read October 16, 2026.
- live read Internet Archive, Wayback Machine copy of mirror.co.uk/robots.txt, October 2, 2024. Supports the date by which both OpenAI tokens were blocked; the January 1, 2024 copy names neither. Read October 16, 2026.