AI creates, the company responds: AI Act transparency obligations

AI does. The company is responding.
The companies are starting a nervous check today:
Does this text have to be marked?
Does this artwork have to be described?
Does the recipient need to know we used AI?

These are important questions.
But not most important.

As of 2 August 2026, new transparency obligations under the AI Act began to apply — including information on the interaction with the AI system and the labelling of specific content generated or modified by the AI.

AI Omnibus, adopted under the Digital Omnibus package, extended the deadlines for the application of some of the requirements for high risk systems to 2 December 2027 and 2 August 2028.

However, it has not deferred the transparency obligations in force since 2 August 2026. And he didn't shift responsibility from organization to technology.

Because even a properly labelled content may contain untruth, violate other people's laws, reveal data, mislead the client or cause a reputational crisis.

And then AI won't sign the explanation.

He won't answer to the client.
He won't stand before the board.
There will be no consequences of the wrong decision.

Responsibility remains on the side of man and organization.

The practice unfortunately shows that many companies have already started using AI before they began to understand the scale of responsibility involved.

Therefore, the real boundary does not exist today between marked and unmarked content.

It runs between companies that can identify the owner of the process and those in which AI is already everywhere, but the responsibility is nowhere.

If the marketing department uses AI to create communication, who approves the result?

If HR supports AI in assessing candidates, who checks whether the system leads to incorrect or discriminatory conclusions?

If a trader generates an offer, who is responsible for the declarations it contains?

If the board receives an analysis prepared with AI, who confirms that the data and requests are correct?

AI Act requires transparency with the recipient.

However, a mature organization also needs internal transparency:
who decides,
who verifies,
who approves,
who reacts when the system is wrong.

The biggest risk is not the lack of a label "generated with AI".

The biggest risk is the lack of a person who can say:
"I am responsible for this process."

Remember: before the company starts to signify AI content, it must first know who is responsible.

#IZOB

#AIAct

#DigitalOmnibus

#AIGovernance

#ResponsibleAI

#Cinnomatech

Source: LinkedIn — IZOB