Particle.news

UK Ruling Recasts AI Liability as Amazon Targets Perplexity and India Weighs Labeling

The ruling redirects compliance toward watermarking, provenance documentation, safer output controls.

Overview

  • The UK High Court held Stable Diffusion’s learned weights are not copies under the CDPA, so importing, possessing or distributing the model in Britain does not constitute secondary copyright infringement.
  • Judges found only limited, historic trademark exposure from Getty/iStock watermarks appearing in earlier outputs, noting that filtering and hosted-platform safeguards have reduced the risk.
  • Getty accepted that training and development occurred outside the UK, weakening its primary claim and underscoring the importance of where training, storage and hosting take place.
  • Legal guidance urges rights-holders to watermark assets, monitor outputs and use contractual limits on AI training, while developers maintain auditable records and robust filtering pipelines.
  • In a separate commercial dispute, Amazon filed a U.S. suit alleging Perplexity’s agentic shopper bypasses its interface and ad units, as India’s proposal for visible synthetic-media labels draws creator concerns and experts foresee more M&E litigation.