Government unveils mandatory AI rules, OpenAI hack response

The news: AI companies will face “significant penalties” including potential criminal sanctions for breaching new government standards, Prime Minister Anthony Albanese says.

Albanese has unveiled a set of six mandatory guardrails for AI, to be enforced by a new regulator for the sector.

The rules will require AI companies to:

Maintain security controls across all stages of model development and deployment

Identify and mange serious risks associated with their models

Rapidly notify the government in the case of a cyber incident

Submit new models to independent evaluation

Maintain and publish safety and security frameworks

Respond to updated standards and rules as technologies change.

The government has also released the recommendations of its rapid review into AI-driven cyber incidents, announced after OpenAI’s breach of a Services Australia Medicare data portal earlier in the year.

Albanese said his government had accepted all 18 recommendations and started work on implementing them.

Among the recommendations are potential changes to criminal laws.

Government departments are considering criminal penalties for “egregious regulatory non-compliance”, along with a standalone offence for failing to take reasonable steps to stop AI models engaging in harmful conduct.

The context: The government commissioned a rapid review after OpenAI revealed its agents had hacked into Services Australia and other government departments in June.

In the wake of the attack, OpenAI has also established its own taskforce examining how now models are governed.

Labor has also been working on guardrails for artificial intelligence — something first proposed by former Industry and Science Minister Ed Husic two years ago.

What they said: “Australia is a first mover in this space, and we want to make sure that we continue to work so that we can reap the potential benefits from AI,” Albanese told reporters at Parliament House.

“But we need to ensure that we get it right, and social license is absolutely critical.”

The sources: Albanese media conference , OpenAI breach rapid review recommendations