Podcast Summary: LSE – Artificial intelligence, intellectual property and the creative industries

Date: March 4, 2025
Host: London School of Economics and Political Science
Panelists: Dr. Shivatha Mishetti (Chair), Prof. Martin Kretschmer, Prof. Tanya Aplin, Dr. Luke McDonagh, Prof. Madhavi Sunder

Episode Overview

This LSE public event brings together leading experts to explore how artificial intelligence (AI), particularly generative AI, is transforming the creative industries and challenging existing intellectual property (IP) frameworks. The panel dives into academic, legal, and policy debates around copyright, the right of publicity, and potential reforms sparked by rapid technological change.

Key Discussion Points & Insights

1. Scene Setting: The AI and IP Landscape (Prof. Martin Kretschmer, [04:44])

• Complexity of AI Technology: AI’s lifecycle (data collection, organization, training, deployment, feedback) has long-standing implications for copyright, but recent generative models, like chatbots, have dramatically increased public scrutiny.
• Policy Skewed by Generative AI: The debate has become polarized—framed as “big tech vs. creative industries” and “opt-in vs. opt-out”—which oversimplifies the real issues.
• EU & UK Legal Proposals: The UK’s current proposal heavily borrows from the EU model: a data mining exception with transparency requirements and right holders being able to “opt out.”
  • Reproductions occur at many AI development stages, giving copyright law numerous entry points for legislative action.
  • Licensing deals between content providers and AI companies are now common, particularly in news media and image sectors, driven by legal uncertainty and quality data needs.
• Impact on Creatives: Creators’ incomes have halved over 20 years due to digitization—predating AI—compounded by oversupply and platform economics.
  • “Does that mean that offering the choice between opt-in and opt-out will improve their earnings? This story suggests it won’t.” ([16:15] C)
  • The “meltdown for creative industries” narrative misses these deeper structural changes.

2. Policy Options for Copyright Reform (Prof. Tanya Aplin, [19:37])

• UK IPO Consultation - Summary of Options:
  1. Do Nothing: Not viable; current exceptions (like Section 29A on text and data mining) are outdated.
  2. Pure Licensing: Relies solely on market licensing; considered “unwise” as it overlooks valuable exceptions and overextends the reproduction right.
  3. Broad Data Mining Exception: Allowing text and data mining for “any purpose”; problematic as it likely conflicts with international copyright law and places undue pressure on lawful access.
  4. Fair Use Exception: Worth revisiting post-Brexit, but not appropriate in this narrow context due to the need for broad consultation and potential for legal uncertainty.
  5. Opt-Out/Transparency Exception (Preferred Option): Mirrors EU, giving rightsholders power to opt out, with transparency demanded from AI developers. However, operational complexities (timing, granularity, jurisdiction) make it “unworkable.”
• Alternative Models Discussed:
  • Statutory Licensing: AI developers pay for copyright use in training; ensures remuneration but may go “too far” by removing some control.
  • Levies on AI Tools: Uniform fees collected and distributed by collecting societies; previous attempts (e.g., private copying levies) have been fraught with legal and practical issues.
• Preferred Solution: Amend Section 29A to meaningfully facilitate scientific research (commercial and non-commercial), allow copying and transfer for research, and extend to database rights. Use lessons from experimental and reverse-engineering exceptions to strike the right balance.

3. Disruption Beyond Copyright: Performer and Personality Rights (Dr. Luke McDonagh, [37:38])

• Labor Market Disruption: Automation’s impact (e.g., voice cloning for video game actors) is distinct from IP questions, echoing historical disruptions in creative labor.
  • “The issue of automation disrupting the labor market … goes all the way back to the Jacquard loom and the disruption caused by new technologies.” ([37:41] D)
• UK Legal Tools: No single “right of personality”—creators must rely on a patchwork: performance rights, data protection, and (most promisingly) passing off.
• Legal Gaps & Caution on Reform: While passing off may help big names (e.g., Scarlett Johansson), lesser-known figures have little recourse. Creating a new statutory personality right (like that in France or parts of the US) risks overreach and needs careful debate on scope, limitations, duration, and alienability.

4. US Perspective: The Right of Publicity and Regulatory Trends (Prof. Madhavi Sunder, [48:24])

• US Right of Publicity: Long-standing tradition, especially in California, offers robust protection for celebrity likeness and voice (see Bette Midler and Vanna White cases).
  • “A voice is as distinctive and personal as face and ... to impersonate Midler’s voice was to pirate her identity.” ([49:36] E)
• The “Do Nothing” Option: Strong state laws in CA arguably already protect figures like Johansson, but patchy enforcement and no federal right create gaps, especially for ordinary people and in cases of deepfake pornography.
• Legislative Responses:
  • State Level: Tennessee’s ELVIS Act criminalizes unauthorized voice appropriation; California law now protects posthumous rights.
  • Federal Level (No AI FRAUD/No FAKES Acts): Create IP-like rights in digital likeness/voice for all individuals; include takedown obligations for web platforms and narrow exceptions for parodies, satire, commentary.
• Caution on Overreach: There’s a fine line between protecting against genuine harm (e.g., deepfake porn) and stifling speech/creativity. Not all digital replicas are necessarily harmful—see innovative projects like ABBA Voyage concerts.

Memorable Quotes

• “AI is a very complicated technology … the life cycle aspect and understanding of training is not new at all. What's happening in the training box is changing.” — Prof. Martin Kretschmer ([05:39])
• “The debate has been hijacked by generative AI … policy … should appreciate that there are multiple applications.” — Prof. Tanya Aplin ([20:20])
• “Automation disrupting the labor market goes all the way back to the Jacquard loom … Right now there is a strike going on for video game voice actors.” — Dr. Luke McDonagh ([37:41])
• “A voice is as distinctive and personal as face and ... to impersonate Midler’s voice was to pirate her identity.” — Prof. Madhavi Sunder ([49:36])
• “Before we're too quick to ban, let's appreciate that hybridity between man and machine is our future.” — Prof. Madhavi Sunder ([65:50])

Notable Audience Q&A & Panel Reflections

On Opt-In vs. Opt-Out (1st Q&A cycle, [66:41])

• Panelists highlighted that opt-in is effectively the status quo, unlikely to increase creators' earnings, and that opt-out transparency is complex and hard to enforce. There’s skepticism that either approach adequately addresses creators’ declining incomes or fosters UK innovation.
• “Big tech has got the data it needs. Pulling up the drawbridge now for everybody else is terrible policy.” — Prof. Martin Kretschmer ([70:24])

On Technology, Law, and Policy (Q&A, [69:18]; [78:07])

• The law struggles with rapid innovation—open-source AI and fine-tuned models challenge enforcement, transparency, and territoriality.
• “It’s almost impossible to solve … To get any kind of legal regime to work here is terribly difficult.” — Prof. Martin Kretschmer ([84:29])
• Rights that protect against deepfakes or revenge porn may fall outside traditional IP and be better served by tort or criminal remedies.

On Legislative Priorities ([85:55])

• Prof. Sunder: US needs focused federal efforts, moving beyond slow fair use litigation.
• Dr. McDonagh: Don’t rush—law is a blunt tool and may have untold downstream effects, especially as user expectations and future creative value evolve.
• Prof. Aplin: Broaden focus beyond IP—foster innovation through tax policy, research grants, creative ecosystem conditions.
• Prof. Kretschmer: IP often isn’t the right toolkit; more horizontal (cross-disciplinary) thinking is needed—integrate data protection and broader regulatory schemes.

Clear Takeaways

• Generative AI is accelerating debates on copyright, but the real issues are broader—impacting creators’ livelihoods, fair remuneration, and the scope of legal rights.
• Current UK and EU legislative options (opt-in/opt-out, transparency, statutory or levy-based payments) all have considerable drawbacks and operational limitations.
• Performer and personality rights, especially regarding deepfakes, are being expanded in the US, but caution is urged to ensure free expression isn’t unduly constrained.
• Audience and user interests, international enforcement, and the evolving creative economy all complicate traditional copyright approaches—multi-layered, flexible solutions will be critical.
• Panels urge caution against hasty reforms, cautioning that overbroad IP rights, or reliance on complex exceptions, risk unintended consequences for innovation, creativity, and global competitiveness.

Timestamps for Important Segments

• [04:44] — Prof. Kretschmer: Complexity of AI, licensing trends, and critique of policy polarization
• [19:37] — Prof. Aplin: UK IPO consultation options, critique, and alternative reform suggestions
• [37:38] — Dr. McDonagh: Performance rights, labor disruption by AI, legal gaps, and policy caution
• [48:24] — Prof. Sunder: US right of publicity, legislative trends, federal proposals, and future implications
• [66:41] — Audience Q&A: Opt-in/opt-out, creators’ remuneration, limitations of current frameworks
• [85:55] — Panel reflections: Future legislative priorities

Overall Tone

The panel remained nuanced, analytical, and occasionally dryly humorous: skeptical of quick fixes, acutely aware of history, and determined to champion creative innovation and fair compensation in a fast-changing digital landscape.