You can't regulate AI you can't see, regulator warns
Visibility into how AI systems work is essential before Australia can effectively regulate or shut down dangerous applications.
Australia is debating how to control risky artificial intelligence. This includes emergency shutdown powers.
Grace Carter, government affairs counsel at Elastic, says visibility matters first. "You cannot regulate what you can't see," Carter said at Parliament House last week.
The proposed AI Kill Switch and Data Centre Control Bill was introduced to federal Parliament this month. For high-risk uses like healthcare, credit or social security choices, organisations need continuous monitoring.
They need to understand how the AI reached its conclusion. Carter compares this to aircraft black boxes, which record what happened.
Investigators can then reconstruct incidents. The risk-based approach matters: low-risk applications don't need the same oversight as systems making critical choices.
- Grace Carter, Elastic
- Speaker
- Before House of Representatives
- Status
Why it mattersWithout visibility, regulators cannot catch problems or enforce accountability when AI systems cause harm.
AustraliaAustralian regulators cannot enforce the proposed AI Kill Switch or other controls without clear visibility into how AI systems make decisions affecting Australians.
✓ Claims checked against the source and corrected before publish. checked 16 h ago
Open this story in InSnip →
