Update (2 August 2026)
When I wrote this article, I didn’t expect the topic of AI memoryisation to come back to me so quickly — and quite unexpectedly. While browsing LinkedIn, I came across an article by Professor Ryszard Markiewicz entitled “Against Copyright Protection for the So-Called Memoryisation of Works”, which addresses exactly the issue at the heart of the Munich judgment.
The title alone suggests that Professor Markiewicz takes a different view from the Munich court. That is precisely why I intend to read the article again, carefully and without rushing. Even after my first reading, however, I can already recommend it to anyone interested in the future of copyright law in the age of artificial intelligence.
For now, I have only had the chance to skim through it. And I already have a few questions. I’m not convinced I agree with every argument. But perhaps that is exactly the point of publications like this — not to immediately decide who is right, but to take the time to ask our own questions.
You can find Professor Markiewicz’s article here:
Against Copyright Protection for the So-Called Memoryisation of Works
(available to LEX subscribers)
I’d genuinely love to hear your opinion. Do you agree with the arguments on AI memoryisation, or do you share some of my doubts?
Feel free to contact me: acybulka@poraj.com
2 August 2026: Mandatory AI Content Disclosure. What do you actually need to change in your business?
Many businesses assume that the AI Act only concerns Big Tech companies such as ChatGPT, Midjourney or Sora. In reality, from 2 August 2026, new transparency obligations will also apply to your business if you use AI to create images, videos, audio recordings or certain types of written content.
This is not another regulation that can be postponed. The deadline is just around the corner.
What exactly is changing?
The legal basis is Article 50 of Regulation (EU) 2024/1689 (the AI Act), specifically paragraph 4, which sets transparency obligations for deployers of AI systems (that is, companies and individuals using AI systems, not only those developing them).
Important: While the Digital Omnibus package postponed some AI Act obligations—primarily those concerning high-risk AI systems—it did not postpone the transparency requirements under Article 50. These obligations still apply from 2 August 2026.
When is disclosure required?
1. Deepfakes
Images, videos or audio that have been generated or manipulated using AI and could reasonably be mistaken for authentic content must clearly disclose that they were artificially created or altered.
Learn more about deepfakes and brand impersonation →
2. Texts concerning matters of public interest
If you publish AI-generated or AI-modified text intended to inform the public about matters of public interest, it must also be appropriately disclosed.
Who is exempt? An important exception
The AI Act provides an important exemption. Disclosure is generally not required for content that has undergone editorial review and control, provided that a natural or legal person assumes editorial responsibility for the published content.
In practice, this means that a standard blog article or product description prepared with the assistance of AI, but reviewed, edited and approved by a human editor, does not automatically require AI disclosure solely because AI was involved in its preparation.
A second exception applies to artistic, satirical and fictional works, where disclosure should be provided in a manner that does not interfere with the enjoyment of the work.
Three practical business examples
To make the rules easier to understand, here are three everyday business scenarios.
Marketing agency
A marketing agency creates a realistic AI-generated video featuring a "satisfied customer" who does not actually exist. This is a classic deepfake that viewers could mistake for genuine footage. Disclosure is required.
Online store
An online retailer creates AI-generated product visuals and marketing banners that are clearly promotional graphics and unlikely to be mistaken for authentic photographs. In this case, the obligation will generally not arise—but each situation must be assessed individually.
Service provider
A service company publishes an expert blog article drafted with ChatGPT but reviewed, edited and approved by its owner, who assumes editorial responsibility. The editorial exemption applies, so automatic AI disclosure is generally not required.
Can you see the difference?
The key question is not "Did I use AI?" but rather "Could my audience reasonably be misled about the authenticity of this content?"
What are the penalties?
The potential sanctions are significant: up to €15 million or 3% of the company's total worldwide annual turnover.
In Poland, supervision is expected to be exercised by the Commission for the Development and Security of Artificial Intelligence, established under the forthcoming national legislation. However, even if the supervisory authority begins operating later, the transparency obligations themselves will still apply from 2 August 2026.
Four practical steps to take before 2 August
1. Conduct an audit
Identify every area in your business where AI is used to generate images, videos, audio or externally published content. AI is often embedded not only in ChatGPT, but also in marketing, customer service and design tools.
2. Determine which content is actually subject to disclosure
Not every AI-assisted image or text requires labelling. The decisive factor is whether the audience could be misled about the authenticity of the content.
3. Introduce an internal procedure
Decide who is responsible for assessing disclosure obligations, how disclosures should be presented and how these decisions will be documented.
4. Train your team
Ensure that everyone involved in creating content—marketing, sales and customer service in particular—understands the new requirements.
Not sure whether these rules apply to your business?
We advise businesses on AI Act compliance and intellectual property protection every day. If you are unsure whether your AI-generated content requires disclosure, get in touch with us.
We will help you identify legal risks, practical obligations and the best way to implement the new requirements—without unnecessary bureaucracy.
We can also begin with an IP audit, providing a clear overview of your intellectual property assets, existing protection and the actions worth taking in the coming weeks.
Sources
📄 AI Act Service Desk – Article 50 (European Commission guidance)
📄 EUR-Lex – Regulation (EU) 2024/1689 (AI Act)
This article is provided for general informational purposes only and does not constitute legal advice. Every situation should be assessed individually. We will be happy to help you evaluate your specific circumstances.
