Follow the latest coverage, related explainers and connected technology stories.
Meta plans to implement an update to its commercial terms on October 30 that would allow it to use content from websites linked to business accounts for access, indexing, analysis, and training, without a clear opt-out option. While the company says the change is limited to consolidating provisions and does not alter how it crawls the web, legal experts warn that it could expand its authority over publishers and web content.
Authors say publishers and literary agents are seeking shares of Anthropic settlement payments even though the rights to some works have expired or they are not entitled to the full amount. Organizations representing publishers and writers argue that part of the problem may stem from inaccurate rights records and a complicated settlement process.
A federal court in Massachusetts ruled that a copyright holder’s belief, even if subjective and unreasonable, may be sufficient to issue a takedown notice under the DMCA. EFF plans to appeal the ruling after the dispute disrupted a citizen-journalism channel shortly before a local election.
The Trump administration submitted a 20-page brief defending OpenAI’s use of copyrighted material without a license to train its models, in the lawsuit filed by The New York Times. The brief does not constitute a judicial ruling, but it could influence the legal debate over whether model training qualifies as fair and transformative use.
U.S. copyright cases do not provide a definitive answer on training AI models on books, as current rulings distinguish between using works for training and copying them from pirated sources or using them to build a directly competing product. The unsettled legal landscape indicates that companies and authors will face a lengthy period of litigation before a stable rule emerges.