· Connectors · 4 min read
AI Data Forces: Board-Level Impact Analysis
External forces significantly impact the data used in AI systems:
- Data protection regulations introduce compliance challenges
- Emerging AI-specific regulations affect data usage and disclosure
- Judicial uncertainty surrounds copyright issues in AI training data
- Contractual restrictions may limit data usage in unexpected ways
Board implications include proactive risk management, strategic planning for regulatory compliance, and careful vendor/partner selection.
Data inputs for AI don’t exist in a vacuum - the source and content of datasets are heavily influenced by a number of external forces, including regulations, court decisions, and contractual obligations. My blog post at 273 Ventures (my legal AI company) explores “Why Data Matters - External Forces in AI” more in-depth, if you’d like to read the full post. Here, I’ve summarized the key points for board-level consideration.
Key External Forces Impacting AI Data Usage
Data Protection Regulations
- Introduce compliance challenges, particularly regarding the right to be forgotten
- Complicate the process of “unlearning” data once a model has been trained
- Emphasize the need for proactive data quality management and careful source selection
Board Consideration: How are we ensuring compliance with data protection regulations in our AI initiatives? What processes do we have in place to manage data subject rights?
Emerging AI-Specific Regulations
- New regulations (e.g., EU AI Act) may impose new obligations on AI development and use
- Potential requirements for disclosure of copyrighted data used in training datasets
- State-level regulations in the U.S. are advancing faster than federal legislation
Board Consideration: How are we monitoring and preparing for upcoming AI regulations? What is our strategy for ensuring compliance across different jurisdictions?
Judicial Uncertainty
- Ongoing legal cases regarding copyright infringement in AI training data
- Lack of clear guidance on how intellectual property rights apply to AI training data
- Varying international approaches (e.g., Japan’s proactive stance vs. U.S. uncertainty)
Board Consideration: What is our risk appetite regarding potential copyright issues in AI training data? How are we mitigating these risks in our AI strategy?
Contractual Restrictions
- Data usage may be limited by terms of service, licensing agreements, or other contracts
- Restrictions can come from data aggregators, distributors, or original sources
- Misalignment between data usage rights and model licensing terms can create legal risks
Board Consideration: How are we ensuring that our AI initiatives comply with all contractual obligations related to data usage? What due diligence processes are in place for reviewing data sources and model licenses?
Strategic Implications for the Board
Proactive Risk Management: The complex landscape of external forces necessitates a proactive approach to risk management in AI initiatives. How can we enhance our risk assessment and mitigation strategies for AI data usage?
Regulatory Compliance Strategy: With evolving regulations, organizations need a flexible and forward-looking compliance strategy. How are we positioning ourselves to adapt quickly to new regulatory requirements?
Data Sourcing and Curation: The source and quality of data are critical for both performance and compliance. How should we prioritize and resource efforts to secure high-quality, low-risk data sources for our AI initiatives?
Vendor and Partner Selection: The risks associated with data usage extend to third-party AI tools and services. What criteria should we use to evaluate and select AI vendors and partners to minimize data-related risks?
Ethical Considerations: Beyond legal compliance, there are ethical implications to consider in AI data usage. How do we ensure our AI data practices align with our organizational values and stakeholder expectations?
Intellectual Property Strategy: The uncertainty around copyright in AI training data may impact our own IP strategy. How should we approach the protection and licensing of our proprietary data assets in the context of AI?
It’s difficult for boards to develop an effective and comprehensive AI strategy without a sense of how their organizations plan to use datasets, models, or AI tools and services. By first understanding the external forces that impact data in AI, boards can better guide their organizations toward responsible and sustainable AI adoption.