Quantexa argues that AI sovereignty is not only an infrastructure debate but a question of control, governance and trust in the data used for high-stakes decisions.
Thomson Reuters states that data entered in CoCounsel is never used to train large language models, that models are accessed only for processing, and that data is protected by contractual commitments and encryption.
BARC's 2026 Data Sovereignty Survey (published May 6) finds that data sovereignty has evolved from a compliance topic into a strategic prerequisite for data- and AI-driven core processes, with legal requirements remaining the dominant external driver (cited by 61% of respondents, down from 69% in 2025 as strategic motivations grow in parallel).
The EU institutions reached provisional political agreement on May 6–13 on the Digital Omnibus on AI, deferring the applicability of high-risk AI system obligations from August 2026 to December 2027 (Annex III systems, including recruitment and certain legal/law enforcement tools) and August 2028 (Annex I regulated products).