Mandatory guardrails on AI

Question

1. In September 2024 the Government released proposals for 10 mandatory guardrails for high-risk AI systems. In December 2025 the National AI Plan abandoned mandatory guardrails in favour of voluntary guidance and the uplift of existing laws. a. Who made the decision not to proceed with mandatory guardrails, and was it considered by Cabinet? b. The Government's own consultation concluded that the current regulatory system is not fit for purpose to respond to the distinct risks AI poses. On what basis did the Government determine that voluntary guidance is an adequate substitute for mandatory requirements? c. The Productivity Commission recommended a pause on economy-wide AI regulation. Does the Government accept that this recommendation concerned economy-wide regulation, not targeted mandatory guardrails for high-risk AI, and how did it distinguish the two in reaching its decision? d. The AI Safety Institute was funded at $29.9 million to conduct a gap analysis of existing laws. When did that gap analysis commence, and when is it expected to be complete? Is the Government satisfied that high-risk AI systems will operate in Australia without mandatory guardrails while it is conducted?

Answer

See attached response.

Documents

• Question with answer (PDF) (opens in a new tab)
• BI-030_Liddle_Mandatory guardrails on AI_DISR.pdf (opens in a new tab)