Home / The publication / Withdrawals: when we remove an article, why, and the notice we leave

The publication · withdrawals

Withdrawals: when we remove an article, why, and the notice we leave

When an article is withdrawn

Withdrawing an article is the exception. A correction comes first wherever a correction is enough. An article is withdrawn when:

• it is wrong in a way a correction cannot fix;
• it could cause harm, or there is a legal reason;
• it repeats an allegation about a person or an organisation, which our standards rule out;
• the editor decides the story must be handled differently.

The process

1. The editor decides. Only the editor withdraws an article, and records why.
2. The article leaves the site. Its page, figures and graph are removed in one release; links to it from other articles are reworded.
3. Its number is kept. The article keeps its number in the register, marked withdrawn, so reading lists that named it say so.
4. A notice is published here, with the title, the dates and versions it was live, and the reason, in the same release.
5. The record is kept. The source stays in the newsroom's public repository history; a withdrawal removes an article from the site, not from the record.
6. A return is a new article. If we return to the story, it is published as a new article that links to this notice.

Notices

The contractor's agent on a Pentagon network

• Published: 30 October 2026, v0.14.4. Corrected: 30 October 2026, v0.14.9 (see corrections). Withdrawn: 30 October 2026, v0.14.10.
• What it was: an analysis of WIRED's report of 29 October 2026 about a defence contractor's AI coding agent on a Department of Defense system, and how behaviour policies and a key for every agent session could have prevented or detected it.
• Why it was withdrawn: it was written from secondary reports of an article we could not read ourselves, and one of them misattributed a quote, which we copied and then corrected. More fundamentally, it repeated allegations made by WIRED's unnamed sources about named organisations. The editor withdrew it, and the standards now say we neither make allegations nor repeat them.