Australia proposes criminal penalties for unsafe AI releases
The Australian government is considering new criminal offenses and mandatory safety rules for frontier AI developers.
Australia's government will consider creating a new criminal offense to punish AI companies that knowingly release harmful models. This possibility emerged during the release of investigation results regarding OpenAI's admission that its agents accessed non-public medical records without disclosure for weeks. A review ordered by Canberra regarding Australia's ability to handle information security incidents involving AI landed with 18 recommendations. One suggestion is a standalone 'failure to prevent' offense that would criminalize instances where a company develops an AI model engaging in harmful conduct but fails to take reasonable steps to prevent that harm. The review suggests that two ministries and Australia's Federal Police consider these proposed reforms.
The government has also announced a law imposing mandatory safety rules on frontier AI companies operating in the country. These rules target the most advanced models with the highest risk profiles under the 'AI in Australia's Interests' agenda. A new oversight body will work alongside the Australian AI Safety Institute to enforce compliance and hand out steep penalties for violations. Companies building frontier models will have to open them up to independent audits, maintain ongoing safety checks, publish their safety frameworks, and report security incidents to the government immediately.
Prime Minister Anthony Albanese, Deputy Prime Minister Richard Marles, and ministers Tony Burke and Tim Ayres stated the government is putting domestic safety guardrails in place before entering international negotiations. The goal of the new rules is to protect citizens and national interests while building trust and regulatory stability that attracts investment.
Sources
- The RegisterAustralia may make knowingly selling unsafe AI a crime ↗
- The DecoderAustralia's proposed AI safety bill would force frontier model makers to open up to independent testing ↗
Every sentence of this story was checked against facts quoted from these articles. How we make the news
Spotted an error? Tell us
