In a recent LinkedIn post, Dan Sherrard-Smith breaks down the implications of new European Union regulations concerning artificial intelligence (AI) content, which are set to take effect on August 2nd. Contrary to a common misconception among founders, Sherrard-Smith clarifies that these rules do not constitute a blanket ban on AI in content creation but rather introduce specific disclosure requirements.
Sherrard-Smith highlights that the EU AI Act’s content disclosure rules target specific scenarios where AI is used to create or manipulate content, particularly concerning realistic depictions of individuals and sensitive topics.
“AI images or videos of a real person doing or saying something they didn’t (deepfakes)”
he notes, is one such area requiring disclosure. He further elaborates on other critical points, including AI voice clones or avatars presented as a founder speaking and photorealistic AI-generated people depicted as real individuals, even if they don’t resemble anyone specific. Additionally, AI-drafted text on news, politics, or public health published without human review falls under these disclosure mandates.
Understanding Disclosure Requirements
According to Dan Sherrard-Smith, the key to compliance lies in understanding when disclosure is mandatory. He points out that the regulations are designed to ensure transparency, especially when AI is used to generate content that could be mistaken for reality or influence public opinion on significant matters.
Sherrard-Smith emphasizes the distinction between AI-assisted content that is reviewed by humans and AI-generated content that is presented without significant human oversight. As he puts it:
“The moment AI generates or manipulates how a real person looks, sounds, or is depicted, add ‘Image/video created with AI’ the first time it appears.”
This clear directive aims to inform audiences when they are interacting with synthetic media or AI-generated information that mimics real-world representations.
When AI Use Does Not Require Disclosure
Conversely, Dan Sherrard-Smith clarifies that not all uses of AI in content creation necessitate disclosure under the new EU rules. He provides several examples of activities that remain permissible without specific AI labeling.
These include:
- AI assisting in drafting a LinkedIn post that is subsequently reviewed and published under the author’s name.
- The use of real photos accompanied by quote cards designed by a human.
- AI-generated illustrations that are clearly graphic or abstract, rather than photorealistic representations of people.
Sherrard-Smith summarizes this distinction with a simple rule of thumb:
“If you review and edit AI-assisted text before it goes out, and you’re not depicting a real person synthetically, you’re fine.”
This suggests that human editorial oversight and the avoidance of synthetic depictions of individuals are key factors in determining disclosure obligations.
Global Reach of EU Regulations
A crucial point highlighted by Dan Sherrard-Smith is the extraterritorial reach of these regulations. He stresses that the rules apply regardless of the creator’s geographical location.
As Dan Sherrard-Smith states:
“It doesn’t matter where you’re based. If your content reaches an EU audience, this applies.”
This means businesses and individuals worldwide must be aware of and comply with the EU AI Act if their content is accessible to audiences within the European Union. He illustrates this with the example of “Mia Zelu,” a virtual influencer who gained significant traction, noting that under the new regulations, the operator of such an account would be required to disclose its AI nature from the outset.
📝 About This Content
This article is based on insights shared by Dan Sherrard-Smith on LinkedIn.
📅 Originally posted on July 30, 2026 | View original post on LinkedIn →