The European Commission’s proposed Digital Omnibus amendments to the GDPR aim to simplify rules for AI training by allowing AI developers to use personal data under a legitimate interest basis with an unconditional opt‑out and introducing a new exemption for residual special‑category data. The author argues these changes miss the mark, creating practical and privacy issues — especially extending opt‑outs to scraped internet data and confusing protections for special‑category information.
