Policy · Client-facing & access-to-justice AI tools

Human escalation and handoff protocol for client-facing AI tools

Protocol listing the triggers on which a client-facing AI tool must stop and hand over to a person (urgency, deadlines, family violence, self-harm risk, vulnerability, conflicts, requests for advice), with response times and records.

Kind
Policy. A policy or procedure for a firm or client.
Where it applies
Australia, New South Wales, Victoria, Queensland, Western Australia
AI use
Yellow AI can help in set ways. A lawyer checks specific points before use.
Law stated as at
17 October 2026
Status
Maintained. Checked against the cited sources on 17 October 2026.
Version
1
Source
Written for this library

When to use it

Adopt before launching any chatbot, intake assistant or self-help tool, build the triggers into the tool's instructions and routing, and test every trigger before launch and after each change.

How AI may be used

AI can help build the escalation matrix from the tool's flows and test whether triggers fire, but the supervising lawyer sets the triggers, response times and safety responses.

AI may

  • From the tool flows, list question types, the point a person must take over and the record that must be kept.
  • Generate test conversations for each trigger, including indirect and non-English phrasing.
  • Summarise escalation logs for the weekly review.

AI must not

  • Must not decide that a risk-of-harm signal does not need escalation.
  • Must not continue a conversation that has hit a safety trigger except to give safety information and connect the person.
  • Must not give legal advice to "resolve" an escalation.

The lawyer checks

  • Every trigger has an owner, a response time and an out-of-hours path.
  • Safety messages and phone numbers are current.
  • Missed escalations found in transcript review are treated as incidents.

Method: how the Green, Yellow and Red levels are decided. This is guidance for planning, not legal advice.

Authorities

Read the full text

More in Client-facing & access-to-justice AI tools

All precedents for Client-facing & access-to-justice AI tools · Tasks and AI use in this area