Rethinking Intellectual Property in the Age of Generative AI: Ethics, Rights and Accountability

Abstract The rapid commercialisation of generative artificial intelligence (AI) has converted long-standing philosophical debates about authorship, property and fairness into live disputes before courts and legislatures. This article examines the ethical dimensions of AI and intellectual property rights (IPR) through a mixed doctrinal–empirical design. It develops a four-axis analytical framework (Consent, Compensation, Credit and Accountability) and tests it against secondary data drawn exclusively from verifiable public sources: the World Intellectual Property Organization (WIPO) patent landscape on generative AI, the Stanford AI Index 2025, curated litigation trackers, court judgments and settlement orders from the United States, the United Kingdom, Germany and India, creator surveys by the Society of Authors, an economic study commissioned by CISAC, and the foundational algorithmic-bias audit by Buolamwini and Gebru. The findings show that (i) generative-AI patenting is highly concentrated, with China-based inventors accounting for roughly 71 per cent of 54,000 patent families filed in 2014–2023, while India accounts for about 2.5 per cent; (ii) United States copyright suits against AI developers more than doubled, from 46 to 101, between June 2025 and May 2026; (iii) the US$1.5 billion Bartz v. Anthropic settlement, at about US$3,100 per work, equals roughly 2 per cent of the statutory ceiling yet is four times the statutory minimum; (iv) harm to creators is concentrated in readily substitutable segments, with 36 per cent of surveyed translators and 26 per cent of illustrators reporting lost work; and (v) the leading global AI-ethics guidelines, although convergent on transparency (73 of 84 documents), do not treat creator consent or remuneration as a distinct principle. Comparative analysis of seven jurisdictions reveals divergent judicial answers to the same ethical question, most recently illustrated by the Delhi High Court's interim refusal in ANI v. OpenAI (24 July 2026) and the Munich Regional Court's contrary reasoning in GEMA v. OpenAI (11 November 2025). The article concludes by proposing a layered governance model that combines training-data transparency, statutory remuneration with collective management, calibrated human-contribution thresholds for authorship, and shared accountability for outputs, with specific attention to the equity interests of emerging economies.