#027 - The Law Is Coming for Your AI Content.
On 2 August, your AI content gets a new audience.
The regulator.
That is when Article 50 of the EU AI Act starts to apply.
Providers of generative AI systems face duties to make synthetic content detectable through machine-readable marks. Professional users can face disclosure duties for deepfakes and for AI-generated text about matters of public interest when there is no substantive human review or editorial control.
The exact duty depends on your role, system, content, and workflow. There are exemptions and a limited grace period for part of the marking obligation.
Your legal team should decide what applies.
Because this is not only a legal problem, it is also an ownership problem.
The label is the last step
Most companies will start by asking:
What label should we add?
That is backwards.
Before the label appears, someone has already selected the model, accepted its sources, edited the output, approved the claim, chosen the audience, and released the content.
AI content is a supply chain.
A model drafts it. An agency edits it. Legal comments. Marketing approves. A platform publishes it.
Every handoff feels reasonable.
Then somebody asks who made the final decision.
Silence.
The label may be visible. The accountability remains invisible.
A grammar check is not human oversight
The most important part of the European Commission's new guidance is not the icon.
It is the standard for human review.
For relevant public-interest text, a superficial grammar or formatting check does not count. Human review means examining the substance with relevant knowledge and professional judgment. Editorial control means having real authority to approve, change, or reject the content, including checking facts and the trustworthiness of sources.
In plain English:
The law is not asking whether a human touched the text.
It is asking whether a human exercised judgment.
That is the Human Premium in regulatory form.
AI can produce the words. It cannot absorb your accountability.
Name one owner
Do not answer this with another committee.
Name one person who owns the path from generated output to public content.
Then give that person a one-page AI Content Accountability Card:
-
Content: What was generated or materially changed, by which system, for which audience?
-
Scope: Are you acting as provider, deployer, or neither? Is the content a deepfake or public-interest text? Does an exemption need legal confirmation?
-
Review: Who examined the substance? Did that person have the knowledge and authority to approve, change, or reject it?
-
Disclosure: What machine-readable mark or visible disclosure is required, where will it appear, and who verifies it?
-
Evidence: Who keeps the source, version, review record, decision, and escalation path?
Five fields.
One workflow.
One name.
Run the card on your highest-risk public-content workflow this week. Not every use of AI is covered in the same way. That is precisely why the decision needs an owner.
A label can tell the public that AI was involved.
Only leadership can say who stands behind the result.
GET THE AI CONTENT ACCOUNTABILITY CARD
Your move.
Operational guidance only, not legal advice. Confirm the application of Article 50 with qualified counsel.
