AI Marking and Labelling: August 2026 Is Closer Than You Think

AI adoption in the UK creative industries is no longer experimental, but operational.

From my lens, the conversation has moved from "𝘩𝘰𝘸 𝘤𝘢𝘯 𝘸𝘦 𝘶𝘴𝘦 𝘈𝘐" to:

• Who owns the responsibility for AI risk?
• Who owns outputs/derivative IP?
• Can liability be pushed down the distribution chain?
• Can tech actually sandbox data from training foundational LLMs?
• Are rights holders actually protected?

Regulation, by contrast, appears to have taken a back seat.

The Grok issue at the beginning of the year sparked debate on Regulators' powers (or lack of) but as always is the case, the industry tends not to wait for a perfect solution but continually carve its own path.

The Government is yet to produce anything substantive following the consultation early last year. The world of AI remains largely unregulated.

However, one regime is moving forward:

The EU AI ACT

𝗔𝗿𝘁𝗶𝗰𝗹𝗲 𝟱𝟬: 𝗧𝗿𝗮𝗻𝘀𝗽𝗮𝗿𝗲𝗻𝗰𝘆 𝗜𝘀 𝗡𝗼𝘁 𝗢𝗽𝘁𝗶𝗼𝗻𝗮𝗹

On 17 December 2025, the EU published a Code of Practice on Transparency of AI-Generated Content (𝘴𝘦𝘦 𝘭𝘪𝘯𝘬 𝘪𝘯 𝘤𝘰𝘮𝘮𝘦𝘯𝘵𝘴).

It is guidance designed to help AI providers and deployers comply with transparency obligations under Art 50 of the EU AI Act.

Under it:

𝗣𝗿𝗼𝘃𝗶𝗱𝗲𝗿𝘀 (developers) must ensure AI-generated or manipulated content is 𝗺𝗮𝗿𝗸𝗲𝗱 in a machine-readable format so it can be detected as AI.

𝗗𝗲𝗽𝗹𝗼𝘆𝗲𝗿𝘀 (users) must 𝗹𝗮𝗯𝗲𝗹 AI-generated or manipulated content where it constitutes a deepfake or relates to matters of public interest.

If you publish synthetic content in the EU, transparency will be a legal obligation.

The Code also encourages a distinction between:

• Fully AI-generated content (no human hand)
• AI-assisted content (a mix of the two)

This distinction matters for governance, contracts and risk allocation.

𝗧𝗵𝗲 𝗜𝗺𝗽𝗹𝗲𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 𝗚𝗮𝗽

The final Code is expected in May or June 2026.

Article 50 is due to take legal effect from August 2026.

That is a narrow window to review:

• AI systems
• Content pipelines
• Contractual frameworks
• Labelling processes

For AV production, digital doubles and synthetic modelling, this is particularly acute.

Transparency will define the next phase of responsible AI deployment.