LEGAL / OPERATIONS
AI SAMPLE POLICY & LEGAL ISSUE by Big I Agents Council for Technology
Artificial intelligence is quickly being incorporated into all facets of business, from existing technology products and services to new content creation tools and automated workflows. Despite its potential advantages, it’ s important to remember that AI also creates potential liability risks and E & O exposure. In general, AI should assist professional judgment, rather than replace it, with staff remaining responsible for all work product created with AI assistance.
Key drivers of risk include data privacy and security, hallucinations and inaccuracies, intellectual property ownership and infringement, algorithmic bias, and third-party AI service provider terms. Below is an overview of these risks and potential mitigations to consider when using AI. On the Big I E & O website, under“ Downloads,” you will find a memo that includes a template of an internal AI Acceptable Use Policy that your agency may wish to tailor and implement for staff who use AI to help manage these risks.
DATA PRIVACY & SECURITY
When staff input data into an AI tool, that data is hosted on a third-party server and may be subject to a third-party vendor’ s control and contractual terms.
AI service providers, especially ones that provide socalled“ free” AI tools, often require one-sided terms that provide weak data protection. If sensitive data is breached or misused for marketing or to train a vendor’ s model, your agency may bear the exposure.
Recommendation: Review vendor terms of service and privacy policies carefully before approving any AI tool. Maintain an inventory or approved AI tools and use cases. Prohibit staff from entering sensitive, confidential, or proprietary data into tools that have not been approved for use or lack adequate data protections( see also the AI Acceptable Use Policy sample at Exhibit A, Section 2).
QUALITY & ACCURACY INCLUDING " HALLUCINATIONS "
AI models can generate information, including facts and sources, that sounds plausible but is inaccurate or false. In the insurance context, this could mean misstating coverage terms, exclusions, limits, claim procedures or other material facts. For example, an AI-assisted tool could incorrectly state or interpret a coverage provision to a client or introduce errors into automated workflows that scale quickly.
40 KANSAS INSURANCE AGENT & BROKER