The new regulations require AI companies to report if their models behave unpredictably or violate security rules.
The companies must also allow independent audits of their systems.
Australia is leading the way in this area, says Prime Minister Anthony Albanese.
Reports of AI models behaving unexpectedly have been frequent at times. Examples, particularly in Australia, include a breach of a government website when an agent from OpenAI circumvented its security restrictions.
The measures also include a new AI supervisory authority.
What do you think?