Colao Defends the DSA and Copyright in Europe's Debate Over AI Regulation

Vittorio Colao, former CEO of Vodafone and now vice chairman of General Atlantic for the EMEA region, spoke at Wave by Vento in Turin on Friday, October 9. Asked whether regulation hinders the development of AI in Europe, he answered: yes, but not every rule equally. He builds his defense around copyright and the Digital Services Act. The available excerpt of the TNW piece contains no detailed arguments, so what follows is only what is known for certain, plus the context in which his words should be read.

What Was Actually Said

Colao's wording is selective. He does not say that European rules work well overall, and he does not say they ruin everything. He separates the rules that hold the industry back from those that do not. The TNW headline records where he places copyright and the DSA: he defends these two regimes. The published excerpt does not name the rules he thinks do more damage. There is no point guessing here: judging his actual list of grievances requires the full recording of the talk.

Why His Opinion Carries Weight

Colao looks at regulation from three positions. He ran Vodafone from 2008 to 2018, so he lived under European telecom regulation. In 2021–2022 he was Italy's Minister for Technological Innovation and Digital Transition in Mario Draghi's government, so he helped shape digital policy himself. Now he is an investor at General Atlantic, where compliance costs directly affect valuations and companies' market entry. That combination is rare, and it is why his distinction between "harmful" and "tolerable" rules is more interesting than the familiar slogan "Europe is strangling innovation."

The Three Regimes at Issue

For orientation, these are the rules that currently set the conditions for AI companies in the EU:

• AI Act: entered into force on August 1, 2024; obligations for general-purpose models apply from August 2, 2025. Fines reach up to 35 million euros or 7% of global turnover.
• DSA: fully applicable since February 17, 2024, with fines of up to 6% of global turnover. It regulates platforms and content, not the models themselves.
• Copyright: Directive 2019/790 permits text and data mining, with rightsholders able to prohibit use (opt-out). The AI Act adds an obligation for providers of general-purpose models to have a copyright compliance policy and to publish a summary of their training data.
• Digital Omnibus: in November 2025 the European Commission proposed postponing some requirements for high-risk systems, which showed that even Brussels acknowledges a problem with timelines.

The debate over whether "Europe is slowing down AI" usually lumps all these layers together. Judging by the headline, Colao pulls them apart. Copyright and the DSA set the rules of the game around data and platforms, and they can be defended as a framework of trust. Requirements for specific systems and deployment deadlines work differently: there, the cost of a mistake is higher for a startup.

What Teams Building AI in Europe Should Do

First: do not treat regulation as a single block. Break it down into copyright, platform obligations, and product risk classification. Each layer has its own owner inside the company and its own deadline.

Second: document the sources of your training data now. The copyright policy and the data summary will be required regardless of how the disputes over postponements end.

Third: if your product distributes generated content through a platform, check whether you fall under the DSA's moderation and transparency obligations. This is not a question only for large platforms.

Fourth: wait for the full recording or transcript of the talk before quoting Colao as a supporter or critic of any specific rule. For now, all that is known is that he defends copyright and the DSA and acknowledges that some rules slow down AI.