To ensure accuracy in its legal AI, LexisNexis unexpectedly hired a large number of lawyers, not just data scientists. These legal experts are crucial for reviewing AI output, identifying errors, and training the models, highlighting the essential role of human domain expertise in specialized AI.

Universal safety filters for "bad content" are insufficient. True AI safety requires defining permissible and non-permissible behaviors specific to the application's unique context, such as a banking use case versus a customer service setting. This moves beyond generic harm categories to business-specific rules.

To manage liability, IP rights, and brand safety, agencies should proactively add 'AI addendums' to client contracts. This practice establishes clear, documented rules of engagement regarding approved tools, data privacy, and comfort levels before any AI-assisted work begins, preventing future conflicts.

The risk of AI unreliability in law is not confined to inexperienced users. Top-tier law firms are also being caught submitting court filings with AI-generated "hallucinations" and fabricated cases. This has resulted in firms being forced to apologize and lawyers on both sides of a case being fined, highlighting a systemic issue.

Despite public hype around powerful consumer AI, many product managers in large companies are forbidden from using them. Strict IT constraints against uploading internal documents to external tools create a significant barrier, slowing adoption until secure, sandboxed enterprise solutions are implemented.

Lawyers are slow to adopt new technologies not just due to personality, but because their professional training and ethical duty prioritize preventing client harm. Doing things the "same way" is a risk mitigation strategy, making any innovation inherently dangerous to them.

While law firms have an inherent conflict with AI due to the billable hour model, the push for adoption is coming from their clients. Corporations are now sending formal requests to their legal counsel, requiring them to use AI tools for efficiency and cost savings, thereby forcing the industry to adapt despite its traditional economic incentives.

VC Keith Rabois highlights a core conflict: law firms billing by the hour are disincentivized from adopting AI that makes associates more efficient, as it reduces revenue. This explains why corporate legal departments are faster adopters—their goal is to cut costs.

When developing AI for sensitive industries like government, anticipate that some customers will be skeptical. Design AI features with clear, non-AI alternatives. This allows you to sell to both "AI excited" and "AI skeptical" jurisdictions, ensuring wider market penetration.