Artificial intelligence is delivering substantial productivity benefits across various sectors, with average net gains ranging from 8.2% to 12.3%. This positive impact is highlighted in a new report from the Morgan Stanley Institute for Sustainable Investing, titled "Governance and AI Adoption." Despite these significant advancements, the report cautions that corporate governance frameworks and model controls are failing to keep pace with rapid AI integration.
Widespread Productivity Gains Identified
The "Governance and AI Adoption" report draws insights from extensive interviews with C-suite professionals and a survey of 200 executives globally who are involved in AI governance. Data collected from interviews with 808 C-suite executives in April 2026, combined with 935 interviews from October 2025, revealed consistent productivity improvements. These gains were observed across 10 diverse industry groups, including banking, software, technology hardware, semiconductors, professional services, transportation, automotive, consumer staples, healthcare, and real estate.
Significantly, the report notes that productivity enhancements were not confined to traditionally tech-heavy industries. Sectors such as consumer staples distribution and retail, which typically have limited board-level representation of AI expertise, also reported notable improvements. Functions benefiting from AI adoption are similarly broad, encompassing customer service, marketing, operations, supply chain management, sales, and human resources.
Governance Frameworks Struggle to Adapt
While the business case for AI becomes increasingly clear, companies are still in the early stages of developing the necessary systems to mitigate AI-related risks. The survey of 200 executives found that nearly half (48%) believe their existing risk-management structures will suffice with some modifications, while 39% are still assessing the applicability of their current frameworks.
A positive trend is that almost 90% of companies have either defined or are in the process of defining where responsibility for AI governance and risk management resides within their organizations. However, the implementation of more specific and granular model controls shows lower adoption rates:
- Only 41% of companies have a formal inventory of AI models with assigned ownership fully in place.
- The same proportion (41%) has established formal testing and validation protocols for AI models.
- Defined standards for AI model design and data use are fully implemented at just 39% of companies.
- A model-risk classification and tiering system is in place for 33% of organizations.
- Structured mechanisms to identify unintended or emerging AI risks have the lowest full implementation rate, at 27%.
Future Risks: Regulation and Execution
The report suggests that these governance challenges are likely to intensify as AI adoption accelerates. Currently, data-related risks—including privacy breaches, cybersecurity threats, and algorithmic discrimination—are the most pressing concern, cited by 56% of executives. However, looking ahead two to three years, the landscape of concerns shifts significantly. Around 35% of executives anticipate regulatory and legal risks will become the biggest challenge, closely followed by execution risks, such as difficulties in realizing expected returns from AI investments, cited by 32%.
Morgan Stanley also highlighted that regulatory complexity stands as the primary impediment to effective AI governance, identified as the main obstacle by 38% of respondents. These findings underscore the urgent need for companies to strengthen their governance frameworks at a pace commensurate with their increasing reliance on artificial intelligence.