Rome, 10 Nov. – (Adnkronos) – “Technological sovereignty does not mean autarky, but governability, capacity for choice and control.” This is the core of the speech given by Engineering CEO, Aldo Bisio, during ‘Sovereign AI for Europe: IS-IA, Engineering Group’s Italian digital architecture’, an event promoted by Engineering and co-hosted at the European Parliament in Brussels by the Vice-Presidents of the European Parliament, Pina Picierno and Antonella Sberna. “Europe was ahead of its time with the AI Act, which immediately focused on security,” Bisio continued. “Now, it is necessary to continue by addressing central themes, such as the transparency of training models and proportionality in the use of artificial intelligence. We, as Engineering, have decided to invest precisely in the finishing touches, that is, those aspects that serve to guide artificial intelligence and govern it.” The analogy used by Bisio is that of a road: “It’s as if the Americans hadn’t put up a guardrail. We, on the other hand, also thought about signage to enable companies and institutions to act on process optimization through artificial intelligence.”
During his speech, the CEO then thoroughly illustrated IS-IA, Italy’s Sovereign Intelligence Architecture, the architectural approach developed by Engineering on its proprietary private GenAI platform EngGPT 2, born with the aim of allowing organizations to use Artificial Intelligence while preserving control over data, models, and knowledge and keeping them within secure, controllable, and transparent perimeters. He emphasized: “The race for increasingly larger and more complex Artificial Intelligence models is not necessarily the path Europe should follow. When we move from training to post-training and specialize models for individual industrial sectors or specific institutional contexts, more agile and flexible models can offer comparable results, requiring far fewer resources. This is the logic we followed in developing IS-IA, an open and modular architecture that allows choosing and orchestrating different models according to specific needs, and EngGPT 2, our generative Artificial Intelligence model.”
Bisio concluded by stating that “this is where I see a great opportunity for Europe: to invest in AI specialization, in the skills and data architectures necessary to concretely bring it into production processes and institutions. At this stage, data control and quality become even more important than the choice of a single model.”