The Australian Government is moving forward with plans to regulate how artificial intelligence (AI) is used in Australia, especially in high risk situations where people’s safety, privacy, or rights might be affected. As it stands, there are no specific laws in place that address AI directly. Instead, the country relies on broader laws like those covering privacy, discrimination, and consumer protection, but these don’t fully address the unique risks AI can bring. Because AI is developing so quickly and can impact many parts of daily life and business, the Government believes the current regulations aren’t enough.

To fix this, they’re introducing two sets of guidelines: voluntary ones businesses can start using now and mandatory rules that will become law in the future for AI that could pose serious risks. The idea is to make AI safer for everyone, encourage innovation, and build public trust.

Voluntary Guidelines

The Government has released voluntary guidelines that any business using AI can adopt immediately. These aren’t laws yet, but they’re designed to help companies prepare for the future. By following these guidelines, businesses can show that they’re using AI responsibly and get a head start on complying with future regulations.

These voluntary guidelines focus on key areas to ensure AI is used safely, such as:

• Testing AI systems: Companies should make sure their AI systems work as intended. This means running tests before AI is launched and keeping an eye on it after it’s in use to make sure nothing goes wrong.
• Transparency: Businesses need to be clear with their customers and users about how AI is being used, such as informing people when decisions affecting them are made by AI.
• Human oversight: Even though AI can do a lot on its own, the guidelines stress the importance of keeping humans in control. This means that there should always be a person who understands the system and can step in if something goes wrong.
• Accountability: Companies should have systems in place to take responsibility for how AI is used. They need to make sure there are strategies for dealing with risks and complying with any legal requirements related to AI.
• Record keeping: To stay compliant, businesses need to keep detailed records about how their AI systems are designed, tested, and used.

Following these voluntary guidelines will help businesses manage AI better now and also make the transition to mandatory rules smoother when they come into effect.

Mandatory Guardrails for High Risk AI

In addition to the voluntary guidelines, the Government is planning to introduce mandatory rules, also called guardrails, for AI systems that are considered “high risk.” These are the types of AI systems that, if something goes wrong, could have serious consequences for people’s rights, health, or safety.

High risk AI could include things like:

• AI used in healthcare to help doctors make decisions about patients.
• AI in law enforcement, which could impact people’s freedom or privacy.
• AI in employment decisions, where the system might decide who gets hired or promoted.

The mandatory rules, once they become law, will require companies to follow stricter guidelines. Some of the key mandatory guardrails include:

• Rigorous testing: AI systems will need to be tested carefully to make sure they perform as expected and don’t cause harm. This includes not just testing the system before it’s used but monitoring it afterward to make sure it continues to function properly.
• Risk management: Businesses will need to identify potential risks that AI systems might cause and take steps to reduce or eliminate those risks. This is especially important when dealing with AI in high-risk areas.
• Accountability and governance: Companies must set up processes that show who is responsible for how AI is used within the business. This might involve training employees to oversee AI systems and putting policies in place to ensure compliance with regulations.
• Transparency with the public and regulators: It will be important for businesses to be open about their AI systems. This includes sharing information about how AI is developed and used with customers, regulators, and other organizations involved in the AI supply chain.
• Human control: Even in high risk settings, AI should not run completely without human intervention. The mandatory rules will likely require a system where human oversight is built in, ensuring that a person can step in if necessary.

Why is Regulation Needed?

The Government’s main goal with these rules is to build trust in AI. Right now, many Australians are concerned about how AI might be used, whether it’s in ways they don’t understand or in situations where mistakes could be costly. Public trust in AI is relatively low because people worry about privacy issues, bias in decision-making, and the possibility of AI making errors that impact their lives.

Additionally, businesses often think they are doing a good job managing AI risks, but studies show that they might not be as prepared as they believe. According to a Responsible AI Index commissioned by the National AI Centre, many Australian companies overestimate their ability to use AI safely and responsibly. While a large percentage of businesses think they’re doing fine, only a small percentage are actually following all of the recommended safety practices. This mismatch is one of the reasons the Government believes stronger regulations are necessary.

By creating clearer rules around AI, the Government hopes to boost public confidence and make it easier for businesses to innovate without fear of causing harm or running into legal issues. This will also help ensure that Australia is keeping up with international standards, as other countries like the European Union and Canada are also moving toward stricter AI regulations.

What Should Businesses Do Now?

Even though the mandatory rules haven’t been finalized yet, businesses that rely on AI should start preparing by following the voluntary guidelines. By doing so, companies will be ahead of the game when the mandatory rules are introduced. These guidelines will not only help with compliance but also improve how businesses manage AI risks across all levels, not just in high-risk situations.

Companies should start by:

• Evaluating their current AI systems: They need to look at how AI is used in their operations and identify any areas where improvements are needed.
• Building transparency: Businesses should start informing their customers about when and how AI is being used, especially when AI makes decisions that impact people.
• Training staff: Employees should be educated about the risks and responsibilities that come with using AI, so they are prepared to manage and monitor the technology properly.

Taking these steps now will help businesses avoid issues later and show that they’re committed to using AI in a safe and responsible way.

Businesses are encouraged to provide their input on the Proposals Paper regarding the mandatory guardrails, with the consultation period open until Friday, 4 October 2024. 

Disclaimer
The material contained in this publication is of a general nature only and it is not, nor is intended to be, legal advice.

🖤💛❤️

AMK Law acknowledges the Traditional Custodians of the land on which we are fortunate to live and work. We pay our respects to Elders, both past and present and further acknowledge the important role that First Nations peoples continue to play within our communities.