AI-generated content is becoming easier to produce. Demonstrating whether content originated from AI is becoming increasingly important.
Around the world, regulators are beginning to require organisations not only to disclose synthetic content, but to make that disclosure visible, consistent, and technically verifiable. For legal and intellectual property firms, professional services, and any organisation whose reputation depends on trust, this is no longer simply a compliance issue. It is becoming part of professional credibility.
From 2 August 2026, the European Union requires certain categories of AI-generated content to be labeled and technically detectable. China has enforced comprehensive disclosure rules since September 2025. South Korea, India, California, and several other jurisdictions have introduced or are implementing similar frameworks.
While the details differ, the direction is remarkably consistent: transparency around synthetic content is rapidly becoming the global expectation.
What does the European Union actually require?
The EU framework distinguishes between those who develop AI systems and those who use AI-generated content.
- Providers of AI systems that generate text, images, audio, or video must embed machine-readable identifiers, such as metadata and watermarking, and provide tools that allow users to verify whether content originated from their systems.
- Users publishing realistic AI-generated media — images, audio, or video depicting people, places, or events, as well as certain AI-generated public-interest text — must disclose that content clearly and visibly when it is presented.
The EU's AI Office has also published a set of official, freely available disclosure icons as part of its Code of Practice on Transparency of AI-Generated Content: AI GENERATED for fully AI-generated content, AI MODIFIED for partially AI-modified content, and a basic AI icon designed to carry an interactive second layer or an alternative label. Organisations may adopt these as a straightforward, uniform way of communicating compliance, or rely on equivalent labels, with no attribution to the Commission or the AI Office required.
Perhaps the most important practical point for professional organisations is this: AI-assisted text published on matters of public interest falls outside the mandatory labeling requirement where it has undergone genuine human review or editorial control and a named person holds editorial responsibility for it. This exception applies to published text, not to deep fakes in image, audio or video form, and it does not affect the separate marking obligations that sit with the providers of AI systems.
To rely on it, the Code expects a firm to identify the person or function holding editorial responsibility by name, role and contact details, to maintain the organisational measures and resources that ensure review takes place before publication, and to make those contact details public. It does not require documenting every individual instance of review. For legal and intellectual property firms, this reframes disclosure as a question of editorial governance: a standing process with a named owner, rather than a label applied article by article. For many such firms, it is among the most useful provisions in the framework, because it rewards the editorial discipline they already practise.
AI content labeling around the world
The European Union is not alone. Countries are moving at different speeds, but almost all are moving in the same direction.
| Jurisdiction | Current position | Key requirement |
|---|---|---|
| European Union | Effective 2 August 2026 | Visible disclosure and technical detectability |
| China | In force | Visible labels and hidden technical markers |
| California | Effective 2 August 2026 | Hidden disclosures and free detection tools |
| South Korea | In force | Disclosure and watermarking obligations |
| India | In force | Continuous visible labels and provenance requirements |
| Latin America | Developing | Brazil advancing; Chile and Mexico progressing through legislative frameworks |
For organisations operating internationally, the practical consequence is straightforward: many firms will comply with the strictest applicable standard long before domestic legislation requires them to do so.
Compliance is only part of the story
The broader issue is trust.
For legal and intellectual property firms, media organisations, and other professional services businesses, every publication contributes to reputation. Being able to demonstrate where content originated, how it was reviewed, and who accepted editorial responsibility is becoming a competitive advantage.
Organisations that approach AI disclosure merely as a regulatory obligation may achieve compliance. Those that approach it as part of their reputation strategy are far more likely to build confidence with clients, colleagues, and the wider market.
The real challenge is process
Adding a disclosure label takes only seconds. Building a reliable editorial process does not.
Organisations will need clear internal policies governing when AI-generated content requires disclosure, how technical markers are preserved across publishing platforms, and how human editorial review is documented.
In practice, workflow — not the label itself — is likely to become the greatest operational challenge.
What to watch between now and 2027
Several developments are likely to shape how organisations approach AI transparency over the coming year.
- The implementation of the European Union's transparency obligations and California's legislation from 2 August 2026.
- The Code sets a hard deadline of 2 February 2027 for providers to make their detection mechanisms interoperable, the point at which cross-provider verification of AI markings is expected to work in practice.
- Adoption of the European Commission's disclosure icons by publishers, platforms, and technology providers.
- Early enforcement decisions that clarify how regulators interpret "clear" and "distinguishable" disclosure.
- Growing reliance on documented human editorial review as both a compliance mechanism and a trust signal.
- Increased scrutiny surrounding synthetic content during major elections and other significant public events.
Common questions about labeling AI content
Do I have to label text written with AI?
Not always. Within the European Union, mandatory disclosure generally applies only to AI-generated public-interest content that has not undergone meaningful human editorial review. Internal documents, private communications, and properly reviewed publications generally fall outside that requirement.
Are the EU disclosure icons mandatory?
No. They are official, freely available resources that provide a simple and recognisable way to communicate compliance. Equivalent disclosures may also be used, provided they are equally clear and understandable.
Does this matter if my firm is outside Europe?
Yes. If your organisation publishes content into jurisdictions where these rules apply, compliance obligations may extend beyond your own country. Many organisations are therefore choosing to adopt a single global standard rather than maintaining different processes for different markets.
What should firms do now?
Review the AI tools currently used within your organisation. Understand which outputs include technical provenance information. Define when visible disclosure is required. Formalise human editorial review with clearly assigned responsibility. These measures not only support compliance — they also strengthen organisational credibility.
Transparency is becoming part of professional credibility
The organisations that adapt most successfully will not necessarily be those that add disclosure labels first.
They will be those that build editorial processes capable of demonstrating how every published piece of content came to exist, who reviewed it, and who ultimately accepted responsibility for it.
As artificial intelligence becomes an increasingly common part of professional communication, transparency is no longer simply about compliance. It is becoming part of trust itself.
If you are reviewing how your firm publishes AI-assisted content and would value an independent strategic perspective, we would be pleased to discuss it with you.