As artificial intelligence continues to expand, subsuming the Age of Information, we as consumers of AI are caught in a mad scramble to first understand it (and it’s potential), and then harness it to our own benefit.
On the 2nd of August 2026, the Article 50 of the AI Act came into force. Who cares what happens in the European Union? Anyone providing services to the 452 million citizens of the European Union, I would imagine.
Once detected, AI-generated assets are getting the down vote:
… a long-established online marketplace for 3D assets used by video game developers, film editors, and 3D printing nerds, users are sending a clear message … the marketplace is being flooded by AI-generated assets, representing one in six models — but they only account for only $1 out of every $90 in revenue.
Here, the covenant between those flooding the marketplace with low-grade AI slop (the Provider), and not labelling their creations as AI-generated, they’re placing the owners of said marketplace (the Deployer) in a bad place from a legal perspective.
I’m reminded of Section 230 of the Communications Decency Act of 1996, and both pieces of legislation revolve around the same question: Who is responsible when a system delivers something harmful, fraudulent, or illegal to an end-user?
| Regulatory Feature | US Social Media Era (Section 230) | EU AI Act Era (Article 50) |
| The Legal Split | Differentiates between the Platform (e.g., Facebook) and the Third-Party User who posted. | Differentiates between the Provider (who built the AI tool) and the Deployer (who sells the output). |
| The Liability Shield | The Platform is entirely immune if a user posts something illegal, provided the platform did not create it. | The Provider is largely shielded from consumer-facing disclosure fines if the Deployer fails to label the content at checkout. |
| The “Active Role” Line | If a platform materially contributes to creating the illegal content, it loses immunity. | If a Deployer significantly alters the AI model, they lose “Deployer” status and are legally reclassified as the “Provider”. |
Do the major AI vendors sit outside the legal protection of Section 230? At the moment, no such ruling has been made, but there are some signals:
- Garcia v. Character Technologies, where a federal court treated chatbot output as a product rather than protected third-party speech — a narrow, non-binding decision.
- Raine v. OpenAI, which tests the same question against OpenAI, is still pre-trial.
You create an unlabelled horrible deepfake, share it on Facebook, and Section 230 swoops in to protect Facebook.
You create an unlabelled low grade digital asset for sale on a digital storefront, and Article 50 ensures no such protection for the storefront.
Here, we see how fragile the covenant between the Provider and the Deployer is, weakened — some (me) would argue — by that implicit thing known as trust, but I’m a cynic. But what mechanisms could we use to enforce appropriate labelling?
As both consumers and the owners of businesses, the Age of Artificial Intelligence is making a difference, no doubt, but are we buying it?