AI and legal tech
AI-agent incident reporting moves into federal inquiry
OpenAI and Anthropic told a parliamentary inquiry they would support mandatory disclosure rules for AI-agent security incidents.
Where it standsParliamentary evidence only. No mandatory AI-agent incident-reporting law has been enacted or introduced in a bill.
What happened
OpenAI and Anthropic told the Joint Select Committee on Artificial Intelligence that they would support Australian laws requiring disclosure of security incidents caused by AI agents. The evidence followed scrutiny of delayed notification after an OpenAI agent accessed Australian government websites.
The committee's terms include AI-related cyber security, national resilience and regulators' ability to respond to emerging risks. Its final report is due on 30 November 2026.
The same hearing exposed competing positions on copyright and AI training. These statements are advocacy and evidence to an inquiry, not changes to Australian copyright law.
Why it matters
Mandatory reporting could affect AI vendors and organisations using autonomous systems. Legal services businesses should not wait for legislation to decide how an agent incident will be detected, escalated and disclosed.
What a firm does with this
For intake and workflow automation, keep an incident register and define who can stop the system, preserve logs, notify the vendor and assess regulator or client notification. Review vendor contracts for reporting timeframes and access to evidence.
Read the source
- Parliament of Australia: Joint Select Committee on Artificial Intelligence
- Reuters: inquiry evidence on incident reporting, 6 October 2026
General information, not legal advice. A summary is not the judgment, bill or release: read the source before you rely on it.