At the recent AI Strategy Summit in New York, Benesch Partner Aslam Rawoof joined a distinguished panel to discuss the legal, ethical and compliance considerations surrounding corporate AI policy. Alongside panelists including Exos General Counsel Marc Mandel and Verizon Managing Associate GC Josh Dubin, Aslam contributed to a rich dialogue on the evolving landscape of AI governance. Here are the key takeaways from their discussion:

Challenges in AI Governance

Employee-Driven AI Adoption:

• Employees, particularly early adopters, are often ahead of formal AI policies. This bottom-up usage necessitates creating user-friendly, concise guidelines for safe AI usage.
• AI’s rapid evolution, especially post-ChatGPT in 2022, complicates policy enforcement. AI tools enter organizations through various channels, making comprehensive policing difficult. Effective governance must integrate policy with technological oversight.

Client Inquiries and Organizational Awareness:

• Many organizations lack awareness of the AI tools in use already by their own employees. Conducting audits to understand tool adoption and use cases is crucial before policy development.
• It’s essential to demystify AI and encourage logical thinking about its use. Employees often have basic questions due to misunderstandings about AI.

Policy Development Strategies:

• Companies should integrate AI into existing policies to avoid contradictions. Initial AI policies should be simple and reviewed frequently (quarterly is recommended) to keep pace with technological advancements. Involving DEI committees can address potential discrimination risks associated with AI.

Addressing Policy Gaps

Employee Data and Confidentiality:

• Policies must cover employee data usage, not just consumer/client data.
• Education is key. Employees need to understand not to share sensitive information with AI tools like ChatGPT, just as they wouldn’t with Dropbox or similar platforms.

Leadership and Education:

• Short, memorable policies and consistent education through meetings and AI-focused channels are effective.
• Use brief webcasts and AI resource channels instead of lengthy training sessions. Leadership involvement is critical for successful AI policy adoption.
• High-level buy-in from leadership ensures organizational adoption and compliance with AI policies. Collaboration between CCOs and CIOs is essential to balance legal and technical aspects.

Monitoring and Liability

Monitoring Tools and Liability Limitation:

• Monitoring tools exist but require human oversight. Their mere presence can be beneficial legally.
• Monitoring should focus on appropriate use rather than punishing AI usage. Embracing AI as a tool with proper parameters is crucial.

Developing Practical AI Policies

• Start with employee surveys and focus groups to understand current AI usage within your organization. Develop policies that are practical and feasible across the organization, involving a broad array of stakeholders.
• Convene an AI committee with representatives from various departments to develop comprehensive policies.
• Understand applicable laws and compliance obligations.
• It’s better to have a good, short policy that can be frequently revised rather than striving for perfection.

Conclusion

The panelists emphasized the importance of a proactive, cross-functional approach to AI policy development. Organizations should reference frameworks like the NIST AI Risk Management Framework, involve diverse stakeholders and continuously adapt their policies to keep pace with AI’s rapid evolution. By fostering innovation and maintaining compliance, companies can navigate the complexities of AI governance effectively.