Fortune Brainstorm AI London conference, May 2025

Key Regulatory Takeaways

• Industry Standards for AI Governance: Lord Clement-Jones proposed that industry-led AI standards, particularly from businesses using AI, could achieve international governance goals without relying on nation-state agreements, which are complicated by U.S.-China tensions and the Trump administration’s anti-regulation stance and threats against countries perceived as targeting U.S. tech firms.

• Technical Standards Needed: Navrina Singh of Credo AI criticized existing standards like ISO 42001 for focusing too much on high-level policies. She emphasized the need for technical standards with specific requirements and testing methodologies to ensure responsible AI implementation.

• Shift to AI Applications: Analyst Benedict Evans noted that as base AI models become commoditized, competition is moving to the application layer. Companies must build specialized applications, integrate with other software, and prioritize user interface design to stand out.

• Custom Benchmarks Required: Thomas Wolf of Hugging Face highlighted that public AI benchmarks are losing relevance as models perform similarly. Companies need tailored benchmarks to evaluate AI performance for specific real-world tasks.