What ethical and legal safeguards should govern how AI companies source training data? Suggest a regulatory framework applicable to India.

Training large AI models depends on vast human-authored text. Reports that leading firms bought rare books, sliced off their spines, scanned the pages and discarded the remains [1] show that data acquisition has outpaced both ethics and copyright law, making safeguards over provenance, consent and accountability urgent.

Ethical safeguards

• Lawful provenance: datasets built only from legitimately acquired material, never from pirate repositories; internal sourcing projects should not stay secret until litigation exposes them [1].
• Transparency and accountability: disclosure of corpus composition, aligning with NITI Aayog's Responsible AI principles of transparency, accountability and equality [2].
• Creator fairness: authors and publishers compensated, since their work is the input to a commercial product.
• Cultural stewardship: digitisation must be non-destructive; irreplaceable editions are public heritage, not disposable feedstock.

Legal safeguards

• Copyright clarity: India's Copyright Act, 1957 offers only a closed list of fair dealing exceptions under Section 52, with no text-and-data-mining provision — leaving AI training legally untested [3].
• Personal data discipline: web-scraped corpora carry personal data, attracting the DPDP Act, 2023 duties of consent, purpose limitation and notice [4].
• Remedies: statutory damages and injunctive relief, so infringement is not merely a cost of doing business.

A regulatory framework for India

• A statutory TDM exception, conditional on lawful acquisition, non-destructive copying and opt-out for rights-holders.
• A dataset provenance register with mandatory disclosure and third-party audit, operationalised through the AI Safety Institute and AI Governance Group created under the India AI Governance Guidelines (2025) [5].
• Collective licensing via copyright societies, creating a royalty pool for authors.
• A national digital deposit so any scanned rare work enters public archives, and the physical copy survives.

India can convert this global controversy into first-mover advantage: a techno-legal regime that keeps data lawful, creators paid and heritage intact will make #AIforAll credible, letting innovation and cultural preservation advance together rather than at each other's cost.