AI regulations diverge globally, putting IP owners under compliance pressure

10 November 2026

As generative artificial intelligence continues to disrupt traditional notions of creativity and innovation, intellectual property professionals face a challenge: balancing technological advancement with legal frameworks designed for human creators. That challenge took centre stage during the session, “The Future of IP in the Age of Generative AI, Autonomous Innovation and Data Privacy”, at the 10th International IP Legal Conference Thailand, 2026 in Bangkok on November 10-11.

The panel examined emerging questions around AI-generated works, inventorship, ownership, data governance and cross-border compliance and how existing IP frameworks are responding to these developments.

Growing divergence in global regulation

Opening the discussion, Bruno Weil, a partner at Weil & Associes, highlighted the fragmented international regulatory landscape.

“We are seeing a fragmented regulatory environment,” he said, noting that major jurisdictions are pursuing markedly different approaches to AI governance, copyright and privacy regulation.

According to Weil, the European Union has adopted one of the world’s most comprehensive regulatory models through the combination of the AI Act, the General Data Protection Regulation (GDPR) and copyright provisions governing training data, including text-and-data-mining opt-outs, alongside transparency requirements for general-purpose AI providers under the AI Act. The United States continues to rely heavily on litigation, judicial interpretation and agency guidance, while China has introduced AI-specific rules covering content labelling, security assessments and platform responsibilities. The UK, Japan, Singapore and India have meanwhile taken their own approaches to AI governance.

These differences have become more apparent over the past two years. The EU AI Act, which entered into force in 2024 and is being implemented in phases, introduced copyright-compliance and training-content summary obligations for general-purpose AI providers in August 2025. Meanwhile, U.S. courts are becoming the primary battleground for AI copyright disputes through lawsuits brought by authors, publishers, visual artists and media organisations against major AI developers.

Weil noted that despite differing legal approaches, he saw a common principle emerging globally. “Most jurisdictions continue to require meaningful human contribution for copyright and patent protection.”

That principle has been reinforced by decisions from the U.S. Copyright Office, which has determined that AI-generated material without sufficient human authorship is not eligible for copyright protection, and by patent authorities in several jurisdictions that have rejected AI systems as inventors, most notably in the widely watched DABUS cases.

Copyright, AI training and remuneration

One of the session’s debates centred on the use of copyrighted works to train AI models.

Cahit Suluk, founding partner of Suluk Intellectual Property Law Firm, framed the issue around three fundamental questions: whether training on protected works infringes copyright, whether AI outputs themselves infringe existing rights, and whether those outputs should qualify for legal protection.

“My position is that AI training should be permitted, but it should be accompanied by equitable remuneration and collective management mechanisms,” he said.

The proposal reflects international debate about licensing and compensation models for AI training. Some collective management organizations representing authors, musicians and publishers in Europe and elsewhere have increasingly advocated remuneration systems that would allow AI developers to continue training models while ensuring creators are compensated.

Suluk added that policymakers should consider remuneration mechanisms when AI-generated content substitutes for human-created works in commercial markets.

“These are policy choices, but protecting human creativity must remain central,” he said.

His comments come at a time when publishers, record labels and collecting societies worldwide are seeking new frameworks for AI licensing. Several major technology companies have already entered content-licensing agreements with news publishers and media organisations, suggesting that commercial solutions may emerge alongside regulatory reforms.

Lucia Batlova | Lead senior counsel @ Lenovo Group

Governance becoming a business imperative

For Lucia Batlova, lead senior counsel for Europe & META privacy and data protection at Lenovo Group, the conversation extended beyond ownership and copyright into practical questions of governance and risk management.

“Companies need to document data provenance and lawful use from end to end,” she said.

Batlova stressed that organisations deploying AI systems must understand how training data was collected, what licences govern its use, and whether privacy obligations, confidential information or trade-secret concerns arise throughout the AI lifecycle.

“We need legal, privacy, security and AI-governance teams involved from the beginning, supported by proper contractual safeguards,” she added.

Her warning reflects a broader trend among multinational corporations. Many organizations are now establishing dedicated AI governance programmes modelled on privacy-compliance structures that emerged following GDPR. Increasingly, boards and general counsel are demanding documentation showing how AI systems are trained, monitored and deployed, particularly when dealing with cross-border data flows and third-party vendors.

Robin Das Gupta | Director of Intellectual Property @ JBT Marel

The ownership question remains unresolved

The session’s most provocative discussion came from Robin Das Gupta, director of intellectual property at JBT Marel, who challenged traditional assumptions about ownership in an AI-driven economy. “Who should own AI-generated works and inventions? The prompter? The employer? The AI provider? Or somebody else?”

Das Gupta also questioned whether prompting alone should be sufficient to satisfy requirements for authorship or inventorship.

“We still don’t have clear answers on whether prompting constitutes meaningful human contribution, or which jurisdiction should govern outputs created through globally distributed systems,” he pointed out.

The issue is becoming increasingly important as AI systems evolve from tools that assist human creators to systems capable of generating sophisticated inventions, software code, designs and creative content with minimal human intervention.

Participants referenced the famous monkey-selfie case, which is commonly cited in support of the principle that authorship requires a human creator. However, panellists acknowledged that generative AI presents a more complex challenge because human involvement often exists on a spectrum rather than being entirely absent.

The jurisdictional challenge is equally complex. AI models may be trained on globally sourced data, operated on cloud servers in multiple countries and used by individuals in other jurisdictions. This raises difficult questions regarding applicable law, infringement claims and ownership rights.

Despite extensive debate, no consensus emerged on how future legal systems should allocate rights when AI contributes substantially to creativity or invention. Nor did participants agree on how remuneration schemes might accommodate patents and technological innovation.

What did emerge was broad agreement that existing intellectual-property frameworks were built around human creators and inventors. As AI systems become increasingly autonomous, lawmakers will be forced to decide whether those principles should evolve or remain foundational to the global innovation ecosystem.

- Darren Barton