LexisNexis AI Insider Webinar: Permissive vs Restrictive AI Regulation

30 Jan 2023

https://www.youtube.com/watch?v=kbHNnmFJS8A

Introduction

• Host: Isabel Cook (Professional Support Lawyer, LexisPlus IP team)
  • Specializes in copyright, database rights, IP transactions, and AI.
• Co-Host: Matthew Newman (Global Chief Correspondent, Emlex)
  • Focuses on data protection, privacy, telecom, cybersecurity, and AI.

Main AI Regulatory Developments

• EU AI Act

  • Comprehensive harmonized rules on AI development and use.
  • Regulation (not directive) - direct application without member state transposition.
  • Political agreement but no official publication; widely leaked.

• United Kingdom

  • No specific AI regulation yet, but political movements include a government white paper "AI Regulation, a Pro-Innovation Approach" and the AI Safety Summit.
  • Various sector-specific regulators are addressing AI impacts.

• United States

  • President Biden's executive order on AI directs evaluation and compliance by government departments and agencies.

Key Themes in AI Regulation

Macro Harms of AI

• Misinformation and Disinformation

  • Misinformation: False/inaccurate information.
  • Disinformation: Deliberately false content for propaganda.
  • Risks: Political disruption, financial manipulation, erosion of trust.

• Privacy and Surveillance

  • Threat to privacy rights due to profiling and data retention.
  • AI may infringe on freedom of expression by erroneous content removal.

• Automated Processing & Bias

  • Risks of biased AI systems in job applications, financial services.
  • GDPR Article 22: Human oversight, explicit consent, necessity for contract.

Intellectual Property Concerns

• Copyright Issues
  • AI's impact on copyright and IP, challenges in IP protection and enforcement.

Panel Introduction

• Keris Swin-Davies (Pinsent Masons)

  • Expert in IP, IT, data privacy, AI, and data commercialization.

• Matt Hervey (Gowling WLG)

  • AI advisor, IP specialist, co-chair of AI subcommittees.

• Guy Matsushita (Nishimura and Asahi)

  • Focus on IP, IT, AI, data; background in engineering.

• Jonathan Armstrong (Cordray)

  • Specializes in technology and compliance, advises on global privacy issues.

AI Regulation by Region

United Kingdom

• No holistic AI regulation.
• Existing laws apply: copyright, data protection, liability, human rights.
• Government's pro-innovation approach; regulatory principles based on OECD.

European Union

• AI Act: Regulates AI; high-risk systems need compliance.
  • High fines for non-compliance, significant risk AI systems.
• AI Liability Directive: Extends product liability to AI.

Japan

• Preference for soft law, guidelines over legislation.
• AI Operator Bylines and risk-based approach.
• Human-centric principles, international standards alignment.

United States

• Permissive approach, exceptions in certain areas like legal tech.
• State and federal AI legislation focus on elections, evidence, transparency.

Key Takeaways

• Training AI Models

  • Jurisdiction matters for copyright exceptions and data protection.
  • Impact assessment crucial for legal and ethical AI use.

• AI Tool Adoption by Businesses

  • Understand AI tools and due diligence on data, confidentiality, liability.
  • Tension exists between AI providers’ secrecy and clients’ transparency needs.

• Governance and Compliance

  • Global consensus on legal, ethical, robust AI.
  • NIST AI risk management framework as a guideline.

• Future Regulation Trends

  • Existing laws (GDPR) will continue to play a role.
  • Possibility of a Brussels effect with AI Act similar to GDPR.
  • Ongoing regulatory scrutiny, necessity for proactive compliance.

Conclusion

• AI regulation is a dynamic and evolving landscape.
• Importance of understanding existing regulations and evolving governance structures.
• Next webinar: "The Future of Law, Grappling with Security and Privacy in Generative AI" at the end of February.