India's AI Copyright Proposal: Online Content Default Training Data, Experts Concerned

• India's draft AI copyright model suggests "lawfully accessed" online content can be used for AI training.
• This implies content is available for AI model training unless platforms implement robust access controls like paywalls or anti-scraping technologies.
• A compulsory licensing mechanism (CRCAT) is proposed, where content accessibility, not explicit permission, determines its availability for AI training.
• Experts warn this model could create a burdensome regime for AI developers due to revenue reporting, government-fixed royalties, and potential for negligible payouts to creators.
• Enforcement for platforms wishing to restrict scraping will depend on their ability to detect and monitor violations.

Why It Matters: India's AI copyright plan may treat online content as AI training data by default.