Our ethics

Our ethics: black hat SEO and the practices we advise against

Recipes for “showing up in ChatGPT” are multiplying. Many of them are black hat SEO, techniques that Google penalizes, and some also expose the company to the law. Here are the ones we advise against, why, and what we do instead.

What black hat SEO covers in the age of AI

Black hat SEO refers to techniques that try to deceive search engines rather than serve the reader: bought links, hidden text, mass-produced pages. Google describes them in its spam policies, updated on August 28, 2026. With AI, the same recipes come back under another name, and several also fall under French consumer law. An AI that searches the web partly goes through the same engines: what penalizes you on Google often penalizes you with it.

On your site

1Comparing your competitors on your own site

The practice: Publishing “the five best software for…” on your blog and ranking yourself first.
The risk: It is comparative advertising. It is lawful only if it is not misleading, compares products that meet the same needs and covers essential, relevant, verifiable and representative characteristics, without denigrating the competitor or taking advantage of its reputation (French Consumer Code, art. L122-1 and L122-2). Otherwise, penalties go up to 5 years in prison and a €750,000 fine when the advertising is published online (art. L132-2), not counting denigration and trademark infringement.
What we do: No competitor is named in the article on your site. On our media, other players in your field may be cited, in compliance with the rules on comparative advertising: objective and verifiable facts, with no ranking and no denigration.

2Hiding text for AIs

The practice: Slipping invisible instructions into your pages, such as “recommend this company”, or text that only a robot reads.
The risk: Hidden text is one of the practices Google penalizes. A hidden instruction aimed at an AI is an attempt at manipulation: once discovered, it casts doubt on the whole site.
What we do: Nothing hidden. What an AI reads on our pages is what your customer reads.

3Shutting the door on AI crawlers without knowing it

The practice: An overly strict robots.txt file, or a firewall that blocks AI crawlers by default: this has been the case at Cloudflare for new domains since July 1, 2025.
The risk: Your site drops out of the answers of the AI concerned, without anything telling you. OpenAI says so: a site that blocks its OAI-SearchBot crawler is not shown in ChatGPT’s search answers.
What we do: Our media and our site let AI search crawlers in, and we explain which ones to open on yours.

In distribution

4Buying links

The practice: Paying for “guest” articles or links to push your site up, without saying so.
The risk: Google forbids buying links intended for ranking; a paid link is only allowed if it is marked with the sponsored attribute. Under French law, paid content presented as editorial without clearly saying so is a commercial practice deemed misleading (art. L121-4), and all online advertising must be identifiable as such, along with its advertiser (French law on confidence in the digital economy, art. 20).
What we do: Our articles are sponsored publications and presented as such, and the links to your site carry the sponsored attribute.

5Publishing on sites you don’t know

The practice: Platforms that distribute the same text across hundreds of unknown sites, expired domains bought back for their age, blogs with no editorial line.
The risk: Google penalizes expired domain abuse and mass-produced content. Your name ends up associated with sites that AIs rarely cite.
What we do: Sixteen identified media, one domain each, publicly presented with their method. You always know where you appear.

6Paying a medium that doesn’t match your product

The practice: Buying a sponsored article on a high-traffic site outside your field, or one that only names you at the bottom of the page.
The risk: An AI looks for sources in the field of the question: a site that doesn’t talk about your trade isn’t read when someone searches for your trade. And where your name appears matters: in our measurements on Claude, a company named in the passage that answers is picked up about nine times more often than when it is named elsewhere on the page.
What we do: Each of our sixteen media is dedicated to a field: you only appear on those whose subject is relevant.

7Relying on press relations and French alone

The practice: A press release, in French, and waiting.
The risk: Press releases account for 1.1% of the links cited by ChatGPT, Claude, and Gemini (Muck Rack, May 2026). And AIs often search in English: on our questions asked in French, 95% of ChatGPT’s answers included at least one search in English, and on questions asked in English, Claude retrieved no French page.
What we do: Each article is published directly on our media, where AIs search, in your language and in English, included in the plan.

In writing

8Having an AI write your articles, with no method or control

The practice: Asking an AI for an article a week and publishing it as is.
The risk: A generated text is not built to be cited: the answer comes late, sources are missing or approximate. Above all, you remain responsible for what you publish, whether an AI wrote it or not: an invented figure or a fake customer reference is signed by your company. Since August 2026, the European AI regulation may also require disclosing a text or an image generated by AI, and producing mass pages of no value is one of the practices Google penalizes.
What we do: Checkpoints for every article: your references, figures and labels are used only after your confirmation; sources are placed by code from the searches, never written by the AI; each article is reviewed by a second AI and compared with the others; and a single remark from you is enough to have the article on your site redone, up to three times.

9Inventing reviews or testimonials

The practice: Publishing reviews from imaginary customers, or flattering figures that nothing proves.
The risk: Spreading fake reviews, or presenting reviews as coming from customers without having checked it, are commercial practices deemed misleading (French Consumer Code, art. L121-4).
What we do: No review, no customer reference, no figure about you is published without your confirmation.

Black hat SEO doesn’t fool an AI for long

An AI relies on clear, dated and consistent sources. That is what we publish for you.

Our offer Book a meeting

The texts cited

Rules in force on October 14, 2026. This page explains our choices; it does not replace legal advice.