The AI Act and the video industry – what you need to know as a content creator or marketer

The AI Act is the first piece of legislation in the European Union to clearly set out how artificial intelligence may (and may not) be used. It will come into force on 1 August 2024 and will apply to anyone who uses AI to create content – including in the video industry.

In short: if you’re creating videos using AI, editing them, personalising adverts or using deepfakes – this regulation applies to you.

In this article, we explain:

• What are the risk levels under the AI Act,

• which uses of AI in video are legal and which are prohibited,

• what responsibilities creators, marketers and agencies have.

What exactly is the AI Act?

This is a new piece of European Union legislation that sets out how artificial intelligence tools should be designed and used – so that they are safe, transparent and do not infringe people’s rights. For the creative industries, particularly the video sector, this entails specific requirements.

You can find the full text of the Act here:
👉 https://eur-lex.europa.eu/eli/reg/2024/1689/oj

How does the AI Act categorise AI systems?

The EU is introducing four risk levels. What is and isn’t permitted – and what your obligations are as a content creator or marketer – depends on the level.

1. Prohibited (unacceptable risk)

Some uses of AI are completely prohibited. These include, amongst others:

• real-time facial recognition on the street (unless it is being carried out by the police under a warrant),

• deepfakes that are not labelled and may be misleading (e.g. political ones),

• AI that detects emotions in schools or businesses,

• the manipulation of children’s perceptions in advertising campaigns.

What does this mean for the video industry?
You cannot create unlabelled deepfakes, impersonate people, manipulate children’s emotions or use biometric data without explicit consent.

2. High risk

This concerns systems that may affect a person’s life or rights – for example:

• AI that analyses candidates on the basis of video recordings (e.g. facial expressions, tone of voice),

• student assessment systems for online courses,

• employee monitoring based on image analysis.

Your responsibilities:
If you use this type of AI, you must keep records, carry out tests, ensure human oversight and notify the system to the EU register.

3. Limited risk

These are common tools which do not define a person, but can influence their decisions. Examples:

• video chatbots,

• AI presenters or voice-over artists,

• personalised video adverts,

• deepfakes labelled as fake.

Requirements?
You must clearly inform the user that the content in question has been generated by AI. Without trying to hide this fact.

4. Minimal risk

These are tools that do not require any paperwork. For example:

• AI for video editing,

• automatic subtitles and translations,

• content recommendation algorithms (e.g. Shorts, TikTok, Reels).

What does this mean in practice?

If you’re making a video:

• Label all AI-generated content – even if it’s just a voice or a face.

• Do not use unlabelled deepfakes – it may be illegal.

• If your tool analyses people (e.g. for recruitment purposes), prepare the necessary documentation and check that it complies with the relevant regulations.

If you’re doing video marketing:

• Make sure that campaigns comply with the law – particularly those aimed at children.

• Only use tools that comply with the AI Act (i.e. those that provide transparency, documentation and labelling).

If you are an agency or a production company:

• Ask AI tool providers whether their solutions comply with the AI Act.

• Educate your customers – you are responsible for ensuring that what you publish is lawful.

• Prepare the team for any audits or queries from customers.

Examples of AI applications in video, categorised by risk level:

How should I prepare?

1. Find out where you use AI in your creative process.

2. Decide which category a given tool belongs to.

3. If necessary, tag the content, gather documentation or switch to a different tool.

4. Choose reputable suppliers who are familiar with the regulations.

5. Establish internal guidelines for working with AI – before anyone asks about legal compliance.

Summary

The AI Act is not just a new obligation, but also a new standard.
It regulates what has, until now, often been done ‘on one’s own initiative’ – namely, the use of AI in content creation.

For the video industry, this means a need for greater accountability, transparency and, at times, simple measures – such as adding a ‘generated by AI’ label.

It is better to implement these principles now than to take a risk later. It is an investment in the trust of your audience and customers.