Require AI companies to publish their training data
Require AI companies to publicly disclose the work their models are trained on.
Signatures
59
signatures
Government response threshold (10,000) · 59/10,000
Debate threshold (100,000) · 59/100,000
- 01 JAN 2026Petition closedFinal total: 59 signatures
- 01 JUL 2025Petition opened for signatures
Background
AI companies may train their models on copyrighted work without a licence to do so, which we think is illegal. However, since training happens in private, rights holders don't know when their works are used, and are therefore not able to enforce their rights. We think training data transparency requirements are urgently needed to combat this use of IP by AI companies.