What Happened
- The European Parliament passed a resolution urging the European Commission to introduce binding rules ensuring that copyrighted content used by generative AI systems is "fairly remunerated" to creators.
- MEPs called for an immediate obligation on generative AI providers — particularly those aggregating press publisher content — to compensate rights-holders rather than simply seeking opt-out agreements.
- The parliament flagged that the copyright-related provisions in the EU AI Act (Regulation 2024/1689), inserted at a late negotiating stage, are too weak and do not meet the demands of the creative sector.
- The resolution reflects the broader global tension: AI companies train on vast corpora of publicly available text and images, including copyrighted work, arguing this falls under fair use or research exceptions.
- The EU Parliament position signals a coming legislative battle between AI developers and creators over the value of training data.
Static Topic Bridges
EU AI Act (Regulation 2024/1689) and Its Copyright Provisions
The EU Artificial Intelligence Act, published on July 12, 2024, entered into force on August 1, 2024, with full applicability from August 2, 2026. It is the world's first comprehensive legal framework classifying AI systems by risk level (unacceptable, high, limited, minimal) and regulating them accordingly.
- General Purpose AI (GPAI) models — including large language models — must comply with copyright transparency obligations: they must document what copyrighted training data was used and maintain policies respecting EU copyright law.
- Free/open-licence GPAI providers have lighter obligations unless they present "systemic risk."
- Copyright-related provisions were added during late-stage negotiations and are widely considered insufficient by creators and publishers.
- Providers must publish summaries of training data but are not currently required to pay rights-holders.
- The Act is risk-based: AI for critical infrastructure, biometric surveillance, or education gets higher scrutiny than a spam filter.
Connection to this news: The EU Parliament's new resolution is precisely a response to the acknowledged weakness of the AI Act's copyright provisions. Parliament is asking the Commission to go further with a dedicated remuneration obligation — separate from the AI Act's transparency requirements.
Intellectual Property and Copyright in the Age of AI
Copyright law grants creators exclusive rights over their original works for a defined period (life of author + 70 years in most jurisdictions). The emergence of large AI training datasets challenges two foundational questions: (1) Does training on copyrighted material without permission constitute infringement? (2) If AI systems extract commercial value from copyrighted work, should creators be compensated?
- "Fair use" (US doctrine) and "text and data mining" exceptions (EU Copyright Directive 2019, Articles 3 & 4) allow limited scraping for research and non-commercial purposes; commercial AI training is contested.
- EU Directive on Copyright in the Digital Single Market (2019): Article 15 grants press publishers a neighbouring right ("press publishers' right") for online use of their content — directly relevant to AI aggregators.
- India's Copyright Act, 1957 does not explicitly address AI training; India's NITI Aayog has discussed an AI governance framework but no binding rules exist yet.
- The fair remuneration debate parallels the music industry's experience: streaming platforms eventually moved from piracy-era free use toward licensing frameworks.
Connection to this news: The EU Parliament is essentially extending the logic of Article 15 (press publishers' right) to AI systems — arguing that just as news aggregators must pay publishers, generative AI that summarises or reproduces copyrighted content should also pay. This debate will directly inform India's forthcoming data governance and AI regulation discussions.
India's Approach to AI Governance
India's approach to AI regulation has been advisory rather than binding. NITI Aayog published the National Strategy for AI in 2018 and Responsible AI principles in 2021. The Ministry of Electronics and IT (MeitY) has taken the position that existing laws (IT Act, Copyright Act, DPDP Act 2023) are sufficient for now and that India should not over-regulate AI at an early stage to avoid stifling innovation.
- India AI Mission (2024): ₹10,372 crore outlay for AI computing infrastructure, AI in governance, skilling.
- IndiaAI platform launched to provide compute access to startups and researchers.
- Copyright Act, 1957: Section 52 lists fair dealing exceptions; training on copyrighted data is not explicitly covered.
- Digital Personal Data Protection Act, 2023: Governs personal data but does not address IP in AI training datasets.
- India's approach: voluntary principles over binding regulation — contrasts sharply with the EU's legislative model.
Connection to this news: As the EU moves toward mandatory remuneration, Indian creators and publishers are watching closely. Indian copyright law's silence on AI training will face increasing pressure, particularly from media houses whose content is already being ingested by global AI systems operating in India.
Key Facts & Data
- EU AI Act: Regulation 2024/1689, in force August 1, 2024, fully applicable August 2, 2026.
- GPAI (General Purpose AI) models must maintain transparency on training data and copyright compliance.
- EU Copyright Directive 2019, Article 15: Press publishers' neighbouring right — relevant to AI aggregation of news.
- Global AI training data market: estimated hundreds of billions of web pages scraped without explicit licensing.
- India's Copyright Act, 1957 has no explicit AI training provision as of 2026.
- NITI Aayog's Responsible AI principles (2021): non-binding voluntary guidelines.
- India AI Mission (2024): ₹10,372 crore for AI infrastructure and governance.
- Life of author + 70 years: standard copyright term in India and EU.