Precisely comments on phase two of the EU AI Act

 

As the next phase of the EU AI Act begins to take effect with the focus on general purpose artificial intelligence (GPAI) systems and requirement for providers to maintain technical documentation, adopt policies which comply with copyright law, publishing detailed summaries of the data that their models are trained on, we talk to Tendü Yoğurtçu, PhD, CTO at Precisely on why meeting the requirements begins with a strong foundation in data integrity.
 
“While the first phase of the EU AI Act encouraged organizations to build awareness around artificial intelligence, this next phase introduces clear accountability.  Providers must demonstrate that their data practices are transparent and aligned with regulatory expectations. To achieve this, organizations need to ensure their data is AI-ready by investing in trusted data foundations that support traceability, accuracy, and compliance at scale.”
 
“This is easier said than done: data often lives in silos and in different systems located across the various functions of an organization.  This leaves data inconsistent, stale, duplicated, and incomplete, which can undermine the reliability of AI outcomes.”

“To support compliance, organizations must break down these silos and have an integrated view of all relevant data across on-premises, cloud, and hybrid environments.  This includes data from legacy systems, especially as critical data often resides on mainframe systems, and requires aligning across business functions.  Doing so improves the reliability of AI systems and helps reduce biases and inaccurate outcomes.  It is equally important to prioritise data governance, quality, and observability to ensure that the data used in AI models is accurate, traceable, and continuously monitored.”
 
“Incorporating trustworthy third-party datasets is no longer optional, it is foundational in the age of agentic AI.  As AI systems increasingly operate autonomously and make context-sensitive decisions, the ability to reason effectively hinges on access to rich and high-quality alternative datasets including location and geospatial data.  Diverse datasets, including demographics, location intelligence, consumer behaviour, and environmental risk factors, can significantly improve the precision and fairness of AI-driven decisions.”
 
“By taking these steps, organizations can build trust in their AI outcomes and demonstrate compliance with the EU AI Act.  Looking ahead, phase three of the EU AI Act, anticipated in August 2026, will introduce strict governance practices, particularly in the public sector.  Investing in strong data integrity strategies today ensures compliance with the EU AI Act while empowering organisations to lead with clarity and confidence into phase three.”