Discussion

Law firms need more than an AI policy to protect client data

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#13077

Law firms using AI need to decide where client information goes, who can access it and what happens if their supplier disappears. Legal Futures contributor Vytenis Sakalas argues that a policy alone will not answer those questions.

Watch Desk analysis

What happened

In an article published on 13 November, Sakalas, CEO of private AI deployment company Moterra, says law firms are already using tools such as Claude for tasks including summarising documents, comparing clauses and drafting letters. He recommends that firms establish how those tools are approved and controlled before clients, insurers or regulators ask.

His proposed setup keeps the AI environment in the firm’s own cloud account, restricts access according to matter permissions, stores logs where the supplier cannot edit them, and requires a named lawyer to approve work reaching a client. He also says firms should consider whether the system could keep running if its supplier stopped trading. Read Sakalas’s article.

Why it matters

These are specific questions about confidentiality, access, accountability and continuity, not simply whether staff have been told to use AI carefully. Sakalas says a direct enterprise subscription can provide commercial terms and administrative controls, but may still leave prompts and logs on the supplier’s platform and access controls separate from a firm’s matter permissions.

That is useful advice, but it comes from a company that sells private AI deployments. Its preferred arrangement is also its business. Firms should treat the article as a vendor’s argument for one model, not as an independent comparison of every available option.

Our read

The strongest point is that a written policy cannot control a tool staff are already using. Start by finding out what is actually in use, then ask who can access client material, where records live and who has authority to approve the output. Less glamorous than an AI strategy deck, and rather more likely to answer a client’s awkward follow-up.

What to watch

  • Whether firms tie AI access to existing matter permissions and information barriers.
  • How they retain records and demonstrate who approved work sent to clients.
  • Whether firms assess supplier continuity as well as the promised protections in a contract.

Discussion spark: Should law firms keep AI systems inside their own cloud accounts, even if that adds cost and complexity, or are strong supplier contracts and controls enough?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.