Verslo žinios covers the approaching EU AI Act transparency deadline of August 2. Our co-founder Gediminas Buivydas explains why the biggest risk for Lithuanian business is not in the new deadlines, but in what has already been law for a year and a half.
Verslo žinios published an article on how a moving target is approaching for Lithuanian business: part of the EU AI Act's provisions on high-risk systems were delayed following the Omnibus agreement, but the transparency requirements for disclosing AI-generated content still take effect on August 2nd. The article features comments from lawyer Rokas Venslauskas and our co-founder Gediminas Buivydas.
Buivydas notes that companies approaching AI specialists still mostly ask what can actually be done with AI and where the value is, not about regulatory compliance. The AI Act typically only comes up as a second step, once a company already has a concrete idea and wants to know whether it's even allowed. He predicts that once the transparency obligations take effect on August 2nd, some of these "what can we do with AI" questions will naturally turn into "how do we do this legally."
In the article, Buivydas identifies a risk that often goes unnoticed: if content labeling becomes too broad and a "made with AI" tag appears at every step, users may simply stop wanting to use AI solutions at all. In that case, the AI label would stop being a mark of transparency and start being a warning sign for a bad experience, and it would be the technology itself that suffers, not just the regulation.
He sees the biggest risk to the AI Act's own reputation in the constant pushing back of deadlines and the lack of clear guidance: businesses are forced to prepare against a target that keeps moving, which reinforces the mistaken belief that the Act restricts competition broadly, when in reality it mainly restricts high-risk use cases specifically.
The core point of the article, echoed by both lawyer Venslauskas and Buivydas: the biggest risk for Lithuanian business isn't in delayed or upcoming requirements, it's in the general AI literacy obligations that have already been mandatory since February 2025. When a company doesn't buy a licensed AI tool, employees use free, personal accounts instead and upload confidential data into them, without weighing hallucination risk or knowing where the line is. That's shadow AI, and its biggest problem, according to Buivydas, is that most businesses don't even see it happening.
Read the full article, with the complete commentary from Venslauskas and Buivydas, on Verslo žinios.
If you want to systematically review how your organization measures up against AI literacy requirements, use our free AI policy generator or book a free 30-minute consultation.