How one contributor got through
The case came to light through an editor outside Canada. On 2 September an editor at Türkiye Today, an English-language Turkish news site, received a pitch from a writer calling himself Daniel Robson, a Canadian independent journalist covering extremism and terrorism, with a headshot and a list of Canadian bylines. According to Türkiye Today's own account, published on 29 September, the writer declined a video call, citing a dental procedure, and its staff later ran his photo through image-analysis tools that flagged it as likely AI-generated. Türkiye Today also alleges a political purpose behind the writing; we have not assessed that claim and do not repeat it here.
On 7 September that editor asked Les Perreaux, editor-in-chief of Policy Options, to vouch for the writer. In a note published on 24 September, Perreaux said he could not. Policy Options had published one article under the name in October 2025, reassured by bylines in other Canadian outlets and by nearly a month of co-operative editing. When it tried to verify him in September 2026, the writer spoke by telephone once but declined a video call, references and a CV; the one Montreal institution he mentioned had no record of him; image-analysis tools raised questions about his photograph; and six unpublished submissions showed signs of AI assistance. He had also declined the honorarium the magazine offers. The writer told Policy Options that threats connected to his work made him unwilling to share personal information, and raised concerns about the editor who had asked the question. Policy Options removed the article and his biography.
The Ottawa Citizen's editor-in-chief, Nicole Feriancek, described a similar pattern on 29 September. The Citizen had published him three times in 2025 and did not pay him; in email he was “cordial, helpful in answering questions, prompt, and deferential”. When the newsroom asked for a call, an in-person meeting and his location, his answers were evasive, and when it said it wanted to verify his identity in person he ended all communication. Feriancek wrote that online tools suggested his headshot and his writing were AI-generated.
Number of articles or submissions, as stated by each outlet
Source: Policy Options editor's note, September 24, 2026; Ottawa Citizen editor's letter, September 29, 2026; both re-read October 10, 2026.
Both outlets have since removed the published pieces. Neither outlet paid the writer.
Perreaux's phrase for what went wrong is the useful one. Each outlet treated the others' publication as evidence, so the bylines vouched for each other. AI, he told The Canadian Press, was “an accelerant”; the core problem was deception.
What the outlets changed
The two outlets responded differently, and the difference matters to anyone pitching them.
The Ottawa Citizen now considers opinion pitches only if the writer is based in Ottawa or frequently spends time there, and agrees to meet one of its staff journalists in person and pose for a photograph. That is a rule about every contributor, not only doubtful ones, and it effectively closes the paper's opinion pages to executives who do not live in or visit Ottawa. Policy Options kept its door open to first-time writers but added a trigger: if a writer lacks a substantial, independently verifiable professional record, or appears to have relied heavily on AI tools, the magazine will take extra steps to establish identity and relevant expertise. Failure to disclose AI use will result in an article not being published.Neither policy mentions agencies or ghostwriters. Both are aimed at establishing that the named author exists and has the background claimed, which is exactly what an editor will ask of an executive whose op-ed arrives through a PR firm.
What Google now says about invented authors
On 1 October Google updated its page on creating helpful, reliable, people-first content. After its existing advice to add accurate bylines, the page now says: “However, avoid using deceptive authorship information. Fabricating creator profiles (such as by using AI-generated headshots, made-up names, or false credentials to make content appear as if it was written by human experts) is a form of deception. Any form of deception makes a page untrustworthy to both users and our automated quality systems, and is a signal of a low-quality page.”
Search Engine Journal, reporting the paragraph on 2 October, noted that Google's quality-rater guidelines already told raters to treat fake creator profiles as deception, and that Google's documentation-updates log recorded a same-day change to its AI-content guidance but not this edit. The page itself is dated 1 October.
Two limits are worth keeping in view. Google does not say how its systems would recognise a fabricated profile. And Google has said before that bylines do not themselves help pages rank; we covered what its documentation does and does not claim about authorship in do author bylines affect credibility signals. What has changed is that the guidance written for site owners now names invented authors as deception. The rest of what Google changed on its two content-guidance pages that day is compared line by line in what Google changed in its content guidance on 1 October.
The other half: sites that invent their journalists
The Robson case is a fake author placed in real publications. The same weeks brought examples of the reverse: publications whose journalists are invented.
The Canadian Press reported on 3 October on We News, a site that labels its “journalist” profiles as AI-generated and, according to an analysis by the detection firm Originality.ai of just over 200 articles under one persona, rewrites stories from outlets including CBC, CTV, Global and the Montreal Gazette. The Canadian Press found its own reporting among the sources. When it searched for the site on Google, it found no results; Google did not say whether it had delisted it, but said its policies prohibit “producing low-value content at scale to game our ranking systems”. In the United States, Futurism reported on 30 September that a group of formerly independent sites it had investigated for AI-written articles under fake expert bylines had stopped publishing and appeared to have been suppressed in Google.
For a company, these sites matter in two ways. Coverage on them is not coverage: a rewritten version of a real article, under an invented byline, adds nothing a buyer or an editor would credit. And a request for comment from a “journalist” who cannot be found anywhere else deserves the same check editors are now applying to contributors. How publications handle mistakes once they find them is covered in how publications handle corrections.
What this means for an executive byline
Neither outlet's policy is about ghostwriting, and an executive who approves an op-ed drafted with help is not a fabricated author. What follows is our reading of what the new checks mean in practice.
Expect to be verified, and make it easy. An editor may now ask for a call or a video meeting with the named author, not the agency. An executive who is available for ten minutes removes the question; one who can only be reached through a press office looks, to an editor primed by this case, a little like the writer who would only use email. Keep the record checkable. Policy Options' trigger is a writer without “a substantial, independently verifiable professional record”. A current company biography, a real profile on LinkedIn and a record of previous coverage are what an editor will look for. Earlier bylines still help, but Perreaux's point about trust cascades means an editor may now look past them to the person. Disclose AI use honestly. Policy Options now declines articles whose AI use was not disclosed. Many outlets have their own rules; asking before submitting is cheaper than a withdrawal. Our explainer on how many expert pitches are written by AI covers how editors are reacting to AI-written submissions more broadly. Use a real photograph. In both Canadian cases, a headshot that analysis tools flagged was among the first warning signs. A recent, real photograph of the executive is now part of the submission, and the Ottawa Citizen will take its own.These checks fall most heavily on contributed opinion, the channel where an outside author writes in a publication's pages. How that channel differs from paid contributor programmes and staff bylines is set out in what a contributor network is.
Methodology and limitations
This guide is based on Policy Options' editor's note of 24 September 2026, the Ottawa Citizen editor's letter of 29 September 2026, Türkiye Today's report of 29 September 2026 and The Canadian Press report published by Global News on 3 October 2026, all read in full on 5 October 2026, and on Google's “Creating helpful, reliable, people-first content” page (last updated 1 October 2026) and Search Engine Journal's report of 2 October 2026, read the same day. Futurism's report of 30 September 2026 was read for its account of the sites' apparent suppression in Google; we did not check those sites ourselves. Four limitations apply. No outlet has established who wrote under the name, and the writer has given Policy Options a reason for refusing verification; we describe what the outlets say they found, not a conclusion about any person. Image and text analysis tools cited by the outlets are probabilistic, and we did not run any ourselves. Türkiye Today's account of the writer's purpose is its own and is not assessed here. And the two newsroom policies are those of two Canadian outlets; we have not surveyed whether other publications have changed their rules.
Conclusion
The Robson case was not mainly about AI. It was about editors accepting other editors' publication as proof of identity, a trust cascade that AI made cheaper to exploit. The responses are concrete: an in-person meeting and photograph at the Ottawa Citizen, identity checks and mandatory AI disclosure at Policy Options, and a paragraph in Google's guidance calling invented authors deception. None of this is aimed at executives who write with help, but all of it raises the cost of an author who cannot be reached. A company placing an executive's op-ed should assume the editor may want to speak to the executive, check the biography and ask about AI, and should have those answers ready before the pitch goes out.
Frequently Asked Questions
Who was Daniel Robson?
The name under which a contributor published opinion articles in several Canadian outlets, including Policy Options and the Ottawa Citizen, in 2025. In September 2026 those outlets said they could not verify the writer's identity or biography and removed the work. The writer declined video calls and other verification, citing safety concerns. No outlet has established who wrote the pieces.
Do newspapers now require op-ed writers to meet in person?
The Ottawa Citizen does. Since 29 September 2026 it considers opinion pitches only from writers based in Ottawa or who spend time there, and who agree to meet a staff journalist in person and be photographed. Policy Options takes extra identity steps when a writer lacks a verifiable record or appears to rely heavily on AI. We have not surveyed other outlets.
Is a ghostwritten executive op-ed a fake byline?
Not in the sense these cases describe. The concern is authors who do not exist or misrepresent who they are. An executive who approves and stands behind an article drafted with help is a real author, though some publications have their own rules about disclosure of assistance and AI use.
What does Google say about fake author profiles?
Since 1 October 2026 Google's helpful-content guidance says that fabricating creator profiles, for example with AI-generated headshots, made-up names or false credentials, is a form of deception that makes a page untrustworthy to users and to its automated quality systems. It does not say how such profiles are detected.
How can a company prove its executive is the real author?
By making the executive available for a call or meeting if the editor asks, keeping a current biography and public professional profile, supplying a real recent photograph, and disclosing any AI assistance according to the publication's rules.
Should I respond to a journalist I cannot find online?
Check first. In September and October 2026 Canadian reporting documented news sites run by AI personas that rewrite other outlets' work. A reporter with no findable bylines at a recognisable outlet, who will not take a call, warrants the same verification editors now apply to contributors.
Sources
- publisher statement Les Perreaux, Policy Options, Why we took down an article published under the name Daniel Robson, 24 September 2026. Read 5 October 2026.
- publisher statement Nicole Feriancek, Ottawa Citizen, Why the Citizen removed articles by Daniel Robson — and what we're doing next, 29 September 2026. Read 5 October 2026.
- press Türkiye Today, Exclusive—Fake AI columnist dupes Canadian press, 29 September 2026. The pitch, the declined video call and the image analysis; its allegations about purpose are not assessed. Read 5 October 2026.
- press The Canadian Press, published by Global News, AI ‘journalists’ start to appear in Canada, 3 October 2026. We News, Originality.ai analysis, Google statement, Perreaux interview. Read 5 October 2026.
- platform documentation Google Search Central, Creating helpful, reliable, people-first content, last updated 1 October 2026. Read 5 October 2026.
- trade press Matt G. Southern, Search Engine Journal, Google Adds Fake Author Warning To Helpful Content Guidance, 2 October 2026; Futurism, That Company That Buys Struggling News Sites and Turns Them Into AI-Powered Content Mills Has Halted Publishing, 30 September 2026. Read 5 October 2026.