Who Are We Actually Keeping Safe?

Hello everyone. Good afternoon. My name is Eunice and I'm from Western Australia. How many of you are here from Western Australia? Nobody. Oh, you? Yes. Okay. That's great. So for those of you who haven't been to Western Australia, I'm going to take you for a tour to Western Australia. Today I'm going to talk about the risk from malicious use, risk management practices and building societal resilience. My title will be, who are we actually keeping safe? And when we gather in this kind of room talking about AI safety, the conversation usually stays very high up.

We discuss and analyse massive data sets from big corporations, institutions, government and departments. But as someone working on the ground at the Western Australian AI Hub, I want to ask you a very basic question. Who are we actually keeping safe? Earlier yesterday, we saw a striking graphic showing the state of AI in 2026. But look closely — what I have added to the screen for most vulnerable people. Our immigrant communities, our seniors and young people and women. Tech safety is not even moving like a turtle.

It is moving at the pace of a snail, left completely behind in the dust, not even under the dust. The UN calls this the evidence challenge. The fact that if we wait for perfect data before we act, the damage in real life is already done. And this is not a theoretical issue. Right now, more than 1 in 5 Australians — over 20% of our population — are completely digitally excluded. Our job is to turn this massive report, [inaudible] a safety report, whatever report that we have, into something that actually can protect them.

So let's look at our seniors. Policy circles tend to treat older people as one single helpless group. But on the ground we see a massive, vibrant spectrum. We have tech enthusiasts who want to build, people who want to learn AI everyday skills and sharp minds who want to debate anything related to ethics, human AI connection, collaboration. They follow the news closely. They are terrified of industrial voice-cloning scams that can copy a grandchild's voice from a public audio clip in a second. 66.5% of all Australians 75 plus are completely left out in the digital conversation.

In cities like Perth and Melbourne, 50% of our older adults were born overseas. English-only safety guidelines simply do not reach them. If we want to build safe systems, they need to be at the table helping us to design the rules, not just receiving brochures or booklets. Next, look at the migrant professionals and families. We are seeing a whole generation of migrant professionals and refugees driven by pure survival. They need a job and income, but they have no network or face significant English as a second language.

AI tools are a massive lifeline for them to understand local workplace, custom and rules. But raw access is not enough. If we don't teach them how to think critically about these tools, we are setting them up to fail. Meanwhile, the kids are caught in a cultural pressure cooker. They are juggling strictly with the parent expectation, and also with the same excellence, with the deep desire to fit into a relaxed Australian culture. In the monitored, unmonitored gap, they use their digital freedom in sneaky, hidden ways. We assume kids are in.

They are not. They are not safe. So, a path forward. We are actually working with the regional organisation as well. And for them, if they do not know how to build AI properly, they will be in trouble as well. So what we do at the Western Australian AI Hub, we are not just doing the academic research. Instead we are giving frontline workers a simple, one-page practical toolkit that people can trust, that is available in the local libraries, neighbourhood centres and multicultural hubs. We want to [inaudible] ask for your attention.

Here in eastern states, you hold all the funding scales and policy level, and we would like your collaboration to work with us in Western Australia. Thank you very much.