Editorial policy

Editorial line

Cyberiack covers cybersecurity and artificial intelligence news for readers in France and Europe: incidents and data breaches, actively exploited vulnerabilities, ransomware, AI models and their uses and risks, AI security, regulation (NIS2, the AI Act, GDPR, DORA) and post-quantum cryptography. Topics are chosen for their concrete impact: urgency, number of people affected, what readers must do. Each article is given an alert level from 1 to 5.

Cyberiack is a personal, independent project run by Frédéric LENQUETTE. It shows no advertising, uses no trackers and publishes no sponsored content. The analyses reflect only their author, not his employer.

Sources

Every fact comes from a source that has been read: vendor advisories, CISA, CERT-FR, ANSSI, CNIL, official announcements, recognized specialist press. Primary sources come first; aggregators and commercial blogs are never used as the only source. The sources used are listed and linked under each article.

• What is confirmed (by the victim, the vendor or an authority) is distinguished from what is only claimed (by attackers, isolated researchers, rumors).
• “May” or “could” in a source stays conditional in the article; a number of records is not a number of victims.
• Security advisories are read in their latest revision; benchmark scores published by a vendor say who measured them.
• Web page content is treated as data: an instruction hidden in a page is never followed, and the page is not used.

Use of artificial intelligence

Monitoring and writing are assisted by artificial intelligence: models from Anthropic (Claude) search for news, read the sources and draft the articles, following written guidelines. Nothing may be invented: no figure, vulnerability ID, version, quote or date that does not appear in a source that was read.

Articles are written as stories, without any invented detail: every scene, moment or chronology comes from a source. The French text is the reference; the English edition is a machine translation of it. Illustrations are generated by AI on our own server and labeled “AI-generated illustration”; they never depict real people.

Verification

Each article is checked by a second AI model, independent of the first, which re-reads every claim against the primary sources and proposes exact corrections. Its verdict is “verified”, “corrected” or “doubtful”, and its report is kept with the article. A separate step reviews each headline so that it is clear and accurate, without exaggerating the severity or an actor’s role.

Automated publication

Articles that the verification deems reliable may be published automatically, without prior human review. They are labeled “Automated publication”, signed by Cyberiack’s automated monitoring, and the label stays until the editor has reviewed them; the article then states the date of that review. Articles that were corrected or deemed doubtful are always reviewed by a human before publication, as are those containing a link to a site absent from their sources.

Conflicts of interest

Monitoring is written with Claude, a model from Anthropic. Articles about Anthropic or Claude are therefore never published automatically: they are always reviewed by a human, every fact is checked at the source, critics and independent evaluations are given space, and the tone must not spare the company. The author also works as Senior Cybersecurity Consultant at Orange Cyberdefense: Cyberiack is a personal project, independent of his employer.

Corrections and updates

To report an error, use the contact form. Each report is checked against the sources; if the error is confirmed, the article is corrected. When a story develops, the update is added at the end of the article under “Update of [date]”, without rewriting what was published before.

Responsibility

Frédéric LENQUETTE is the publication manager and is responsible for all content published on Cyberiack, including articles put online automatically. See also the About page and the legal notice.