Advancing Responsible AI Innovation A Playbook 2025

CASE STUDY 4 Infosys “comply up” general standard Infosys adopts a “comply up” strategy, applying the highest global AI compliance standards – like those in the EU AI Act – across all operations worldwide. This unified approach eliminates complexity from fragmented regulations while exceeding client expectations, as partners increasingly demand robust, responsible AI practices regardless of local requirements. Infosys proved this model’s effectiveness with data privacy, where compliance with the California Consumer Privacy Act (CCPA) created a strong baseline for global operations. Key insight Adopting the highest responsible AI standards across all jurisdictions streamlines operations and ensures consistent compliance regardless of the regulatory regime. Government leaders Actions for government leaders –Resolve regulatory tensions and ambiguities between sectoral and cross-cutting AI regulations: Provide organizations with clear guidance on compliance requirements.35 –Prototype AI governance frameworks: Enhance policy efficacy and feasibility, and mitigate externalities (e.g. economic or rights/freedoms infringements), through policy prototyping, which borrows design and research practices from products and services.36 Best practices include: –Incentivized participation: Across organization size, sector, expertise and the public to ensure prototyping considers all impacted parties and their concerns (e.g. intellectual property loss). –Clear criteria: Set goals, metrics and benchmarks for success upfront. –Robust methods: Avoid testing in isolation from existing policies, policy-making processes and enforcement practicalities. Layer prototyping approaches and prototype at multiple stages (see Figure 4). Refine through agile iteration cycles and feedback loops. –Transparent process: Document and communicate decisions, changes and rationale throughout the process. Provide sufficient time for submission and review of feedback. –Independence: Prototyping should be adopted in a manner that bolsters rather than impedes a policy-making process, representing and benefiting the entire population. For example, organization participation for ulterior motives (e.g. regulatory capture or dilution of policy accountability) should be deterred. –Promote jurisdictional interoperability through multilateral AI governance frameworks: Set shared principles, standards and certification protocols to drive innovation and safety while respecting national interests. Help businesses make sense of multiple frameworks through developing crosswalks e.g. the National Institute of Standards and Technology (NIST) AI Risk Management Framework (RMF) and Japan AI Guidelines for Business.37 Consider participation in multilateral forums that enable international cooperation (e.g. the World Economic Forum’s AI Governance Alliance and the Commonwealth Artificial Intelligence Consortium), as well as collaboratively working towards reducing fragmentation (e.g. China- Pakistan AI Cooperation efforts on innovation and governance).38Key roadblocks organizations encounter from the broader ecosystem Tensions arising from conflicting laws and overlapping authorities, creating difficulties in law enforcement and compliance33 AI regulatory fragmentation and policy instability, generating prohibitive compliance costs34 and threatening confidence in investments in responsible AI practices Advancing Responsible AI Innovation: A Playbook 16