AI Content Governance

How we govern AI-assisted content on this site, and the evidence that the rules actually run.

Build gate status: passing

27 published posts · 4 AI-assisted · 0 policy errors · 0 warnings. This page is generated at build time from the same policy module that blocks the deployment.

The rule we hold ourselves to

Article 50(4) of Regulation (EU) 2024/1689 requires deployers publishing AI-generated text on matters of public interest to disclose it, unless the content underwent human review and a person holds editorial responsibility for its publication. We do not rely on that exemption to stay silent. On Tier 1 content we satisfy the exemption and disclose anyway, because an exemption you cannot demonstrate is not a defence.

Risk tiers

Tier	Scope	Control
Tier 1 Regulated / public interest	Categories: Compliance, GDPR, AI, Business Strategy. Any post tagged eu-ai-act, compliance, gdpr, accessibility.	Named human review mandatory. Fully-generated content prohibited. Visible disclosure always, even where the law would permit silence.
Tier 2 Brand and technical	Product pages, technical walkthroughs, campaign copy.	Human review required. Provenance recorded. Disclosure by channel policy.
Tier 3 Operational	Meta descriptions, alt text drafts, tag suggestions, internal summaries.	Sampled review with tracked defect rate.

Controls that run automatically

Tier is derived from content, not declared. An author cannot lower their own obligations by editing a field; a declared tier may only be stricter than the derived one.
No named reviewer, no publish. AI-assisted content without a human holding editorial responsibility fails the build.
Reviews expire. After 365 days the editorial-responsibility claim is flagged for re-review.
Model identity is recorded, so that "which pages did model X touch?" is answerable after an incident.
Silence is not a claim. Tier 1 posts must declare provenance even when no AI was used.

AI-assisted content register

Every article on this site where generative AI was involved, with the person accountable for it. The full machine-readable record, including human-authored posts, is at /ai-provenance.json .

Article	Role of AI	Model	Editorially responsible
Content Governance Before and After AI: What the EU AI Act Changes	co-authored	`claude-opus-5`	Ahmed Chaabni reviewed 2026-09-07
How We Made Gladtek.com Readable by AI Agents (and Why You Should Too)	drafted	`claude-opus-4.8`	Ahmed Chaabni reviewed 2026-06-30
Building a Jahia Westworld Demo with Agentic AI: From Specs to Video	drafted	`claude-3-5-sonnet`	Ahmed Chaabni reviewed 2026-06-11
The 'AI-First CMS' Myth: Why you should ignore the buzzword	drafted	`claude-3-5-sonnet`	Ahmed Chaabni reviewed 2026-04-16

Article 4: AI literacy

Everyone at Gladtek who operates an AI system on our behalf works to this policy, which is versioned in the same repository as the site and reviewed when it changes. Questions about a specific article, or a correction to the register, go to our contact page .