ALEXANDER, Mr David, Chief, Policy and Advocacy, Australian Chamber of Commerce and Industry CREELMAN, Mr Ross, Associate Director, Digital, Cyber and Energy Policy, Australian Chamber of Commerce and Industry SCHMITKE, Mr Shaun, Head, Workplace Relations Policy, Australian Chamber of Commerce and Industry [by audio link] [15:39] CHAIR: Welcome. I understand that information on parliamentary privilege and the protection of witnesses giving evidence to this inquiry has been provided to you. Thank you very much for appearing before us today. We'll kick off with your short opening statement. Please make it as short as you can so that we have time to get into questions. Mr Alexander : Thank you to the committee for the opportunity to appear today. ACCI is Australia's largest and most representative business network, working through our state and territory chambers and our member national industry associations, representing businesses across the economy—small, medium and large. Our starting point at ACCI is that artificial intelligence represents a major economic opportunity for Australian business. At scale, AI has the potential to lift productivity across the economy, helping businesses automate routine processes, make better use of data, improve customer service, strengthen cybersecurity, accelerate research and innovation and develop new products, services and business models. For Australia, the opportunity is therefore much bigger than simply adopting generative AI tools. It's about how AI can support transformation across established industries, help turn Australian businesses and make them compete internationally and strengthen our capacity to turn world-class research and innovation into commercial outcomes. But capturing those benefits requires broad adoption across the economy, including among small and medium businesses that make up the overwhelming majority of Australian businesses. For many of those businesses, the challenge is no longer simply awareness of AI; it's moving from experimentation towards using AI to build more productive and competitive businesses. That requires practical support to build better digital AI and cyber capability alongside a workforce equipped to adapt as the mix of tasks within jobs changes. We also need to get the regulatory settings right. AI creates new capabilities and new risks, but new technology does not automatically mean existing laws have stopped working. We should start with the specific harm, identify whether there's a genuine regulatory gap and then consider whether additional intervention is necessary and proportionate. Ultimately, Australia should be aiming for responsible AI adoption at scale, supporting productivity, investment, innovation and competitiveness while appropriately managing risk. CHAIR: Thank you, Mr Alexander. Your submission notes that many businesses, in particular SMEs, lack a formal structure and the expertise to fully utilise and adopt AI in order to capture those benefits. What practical supports do businesses need to implement AI effectively while, at the same time, ensuring that that's being done in a safe and transparent way? Mr Alexander : Small businesses clearly lack the capabilities of larger businesses, and they often lack the resources and time to be able to get up to speed. We have to always bear that in mind with this particular cohort. We're very interested to encourage the uptake of AI with these businesses and build their skill levels. That can be done in a whole lot of ways. You've got to start with some of the broad policy settings that the community faces. A lot of the business community face some challenging policy settings on tax, for a start. This does impact on their general ability to compete with peer countries. Australia's company tax rate is 30 per cent, which is the highest among developed countries. This makes them uncompetitive and less able to financially upgrade, because they're lacking the financial resources. We've had some recent challenges with capital gains tax changes and forthcoming trusts changes. These are going to add to the headwinds for Australian business. We suggest that there might be some room to move around the instant asset write-off. The government did make permanent that write-off at budget time, and we commend them for that, but there's still scope to expand the eligibility for that. Thresholds currently at $20,000 could be shifted upwards and a greater allowance made for digital assets to come into that net. That's one more-direct financial aspect, but clearly there are training aspects as well, and perhaps here I could move to our digital chief, Ross Creelman, to expand on some of the aspects of that. Mr Creelman : I would start by saying that I think it's quite clear to all of us that Australian small and medium businesses are characterised by their creativity and ingenuity. The questions we are hearing among our members are no longer, 'Should I be using AI?' and 'How should I be using AI?' Rather, it has become more a question of, 'How can I employ AI to more fundamentally transform the nature of my business to be more competitive and to be more competitive internationally?' David mentioned the question of skills. This is something that is also very much at the front of our mind. We gave evidence earlier in the week to the House inquiry on cybersecurity and SMEs. Something we would say here is that while small and medium businesses are very interested to learn more about how they can use AI and develop those skills, both as a business and as individuals, we maintain the importance of not neglecting the most fundamental skills—cybersecurity et cetera. These are absolutely fundamental before the more exciting AI opportunities are developed. Mr Alexander : I would just add that we're essentially looking to enhance the diffusion of knowledge about AI and how to utilise AI throughout the business community. Some of these are hard to reach, so it's not going to be one simple answer for that massive education task. It's going to be through various existing educational institutions, for example. It's going to be through government. It's even going to be through organisations like ours. We have been progressing our own AI capability uplift programs, and we could go through that if you want. What we bring to it is that we've got a vast national network. We've got something like 430 chambers of commerce in Australia. For members of parliament, that's probably two or three in your own electorates. But what we can do and what we do is that we often run education programs through local chambers, who can bring together businesses and inform them of some of the practical-use cases for AI. This is not the whole solution, but it's part of it. Some of this is about partnering with some of the AI experts to bring their knowledge to the broader business community. These are some of the channels through which we can uplift skills. CHAIR: Are there any examples that you can provide for the committee of where adoption has worked well? We hear about them. We hear the stories, including even in my own electorate. One of the keynote speakers at a function that a local footy club held was a guy talking about the benefits of AI and the adoption for, in particular, small and medium businesses in my area. Clearly we are starting to see more and more businesses take this on and to truly benefit from it. Can you provide some examples of where we think this is working well? Mr Alexander : On notice we could. CHAIR: Sure. Mr Alexander : But, in the broad, it's covering everything, to varying degrees, whether it's managing payroll, marketing, client management or workflow. All these different things are being taken on. But businesses really want what's practical. They don't want to have to spend too much time thinking about it. So: 'What's the best practice in my field? And how can I adopt that?'—or, for some of them, where they can break ground themselves in adopting the technology for their own purposes. It's quite an interesting and exciting story, but we need to be abreast of the world's leading-edge companies here. CHAIR: Are your members reflecting on any concerns around where we do need to regulate around safety in this space? What sort of feedback have you been receiving from your members on that? Mr Alexander : We do have a strong interest in cybersecurity. In fact, the Prime Minister's got a cyber executive council, and we co-chair a cybersecurity-for-small-and-medium-business aspect. In the ecosystem of digital, that cohort has vulnerabilities because it's less well resourced than the larger businesses. AI adds a new dimension to this whole question. For us, part of it is recognising the limitations of business to be able to deal with the threats. Part of it is making them aware of what threats they need to attend to. We're constantly looking for ways to make them less vulnerable to various threats. A lot of the conversation here is centred around the American AI providers. One of the concerns was that the government earlier this year banned DeepSeek from Australian government use because it was presenting unacceptable risks. But there's a question for the business community about the advice that's gone to government agencies and departments about the risks of some of these Chinese models. A number of Australian businesses might be looking at adopting those models because a lot of them are very cheap. Are those models potentially putting Australian businesses at risk inadvertently because they are buying on price but unaware of potential issues that might come that we haven't been informed about in particular? It does seem like the government's informing its own areas, but there's the broader Australian business community to consider as well. CHAIR: Deputy Chair. Senator DUNIAM: Thank you very much to the representatives of ACCI for the evidence. In the opening statement there was reference to regulation. I think it's important, obviously, to get the balance right. There is no need to add another layer of everything to chill any potential benefits that might flow from investment. Of course, I agree. But you referenced identifying any gaps in existing regulation. Has there been any work undertaken by ACCI around what laws exist today that would be useful to basically use as a grounding point for any work government might do in terms of regulation required? And have you identified any gaps that do need filling based on your work so far? Mr Alexander : Our approach is really to ensure that any problems are identified, and not to act precipitately but to ensure that we don't essentially jump at shadows. Let's see whether they really turn out to be serious risks or not. The government hasn't come up with an overall AI act. It's looking to deal with most issues on a general-principles basis. We think that's a fair approach, because a lot of the existing legislation covers the sorts of things that people are concerned about. That's our broad approach. Senator DUNIAM: That's understood. I was hoping you might have had a ready-made plan for us, but I've missed out there. Have you identified anything from counterpart organisations overseas that we should be avoiding when it comes to regulation? Mr Alexander : I think perhaps the EU have become a bit heavy handed in their approach. We haven't looked in detail at other jurisdictions. For us, it's looking at the issues case by case and examining whether there's further action required or not. Senator DUNIAM: Thank you very much. We might stay in touch on that one, then. CHAIR: Ms Chaney. Ms CHANEY: This may move more into the evidence you've given to the other committee. It seems to me, in what I've heard from small businesses, that the AI opportunities seem a few steps ahead of where people are dealing with the day-to-day challenges of running a small business, but the AI risks of cyber, in particular, may catch people unawares. What do you think could be done to better prepare small business for the expected huge leap in cyber capabilities that has the potential to have a real impact on Australian businesses? Mr Alexander : I think the advent of AI has added a new dimension to the existing cybersecurity threats, and this does present a concern for small business. Obviously, with each of these technologies, the barriers to bad actors become lower, but we shouldn't lose sight of the fact that good actors also have the same technology to use for good. We'd want that balance of the good actors being superior to continue. That's the broad approach. For small businesses, it's a matter of being aware of the threats. It means the relevant authorities in Australia being aware and ready to act on any of these developments. The government has got various institutions at work on monitoring various threats, but it's a matter of constant vigilance in this area. We've seen some of the spectacular incidents that have occurred. Below that level, we need to be aware that small businesses, as well, can be vulnerable to these sorts of incidents. Ms CHANEY: It's probably mid-sized businesses that hold datasets that are most at risk. From talking to technical experts about very recent AI developments, it does seem that there's an expectation that there'll be a step change in the ability of agentic AI to find and pursue vulnerabilities. Do you think there is a level of awareness within the business community of how much these risks are changing? Obviously, cybersecurity has always been a threat, but it does feel like there's an order-of-magnitude change in that risk. Is the awareness there? Mr Alexander : I think it's patchy. Parts of the business community are very well informed on these issues, but there's a long tail that is behind the curve. At any point in time, it's incumbent on everyone in the community, from the government and its agencies to bodies like us, to raise awareness and educate the business community on any of those threats and what they can do about them. It's also incumbent on businesses themselves. There are cybersecurity firms that are themselves providing training and operations to service the business community, and you can see that they are having to expand their armoury to deal with the new wave of cyberthreats. Ms CHANEY: Thank you. That's all from me. CHAIR: Senator Sheldon. Senator SHELDON: Thank you, Mr Alexander, Mr Schmitke and Mr Creelman. I just want to go to your submission—and thanks again for the submission and the thoughts that you've presented to the committee. It's helpful in our deliberations. In your submission, you mention that schedule 2.3 of the Fair Work Regulations and health and safety laws are examples of ways AI is already regulated. I will take you to this statement from the ACTU submission, which you may have read: … two Registered Nurses in a public health facility were required to use an AI system to record patient observations. These nurses could not alter the output of the AI system, and the system misrepresented their positions as more senior and highly qualified Nurse Practitioners, which is an offence under Health Practitioner Regulation National Law. This breach was investigated by the Australian Health Practitioner Regulation Authority, putting the nurses' professional registration and careers at risk. Currently, workers don't have an affirmative right to refuse to use unsafe AI technology. Is that an example of where the law is falling short? Mr Alexander : I might refer to Shaun. Shaun, are you online? Mr Schmitke : I am indeed. I have not read the ACTU submission in detail, so I can't really comment on that particular circumstance save and except for saying that whether or not AI was a factor in that particular scenario is not necessarily the big question. The question relates to the conduct and the impact on workers and whether or not those particular workers have some sort of recourse through the existing framework that applies to them or the Fair Work framework generally. Senator SHELDON: Let's go to the SDA submission then. It contains a number of examples where the law is falling short of protecting workers from AI. It says a cashier working at one of the big supermarkets, with 30 years experience, recently received a final written warning because the AI monitoring system had flagged her for making too many errors. At no point was she ever told that the error rate was acceptable, what error rate would trigger disciplinary action or whether a human verified the decision the AI made or reviewed the flags before she was disciplined. The error rate that got her in all the trouble was 0.18 per cent. That means she was making roughly one error per 533 items scanned. Is that a use of AI that you think is socially acceptable? Mr Schmitke : The question of whether or not it's socially acceptable isn't one I, particularly, can comment on in that circumstance, but what I would say is that that would represent, in my experience, grounds for that employee to notify a dispute with their workplace and seek assistance, whether it's of a union or somebody else, to resolve that dispute and challenge the findings or the disciplinary action that has been levelled against that employee. If that employee has been terminated, I would suggest that the Fair Work Act has numerous grounds currently available to challenge a termination in those circumstances on the basis that there's a clear procedural deficiency leading up to the termination of that employee. Of course, we've got general protections and adverse-action provisions under the Fair Work Act which would also be available to that employee. Senator SHELDON: Should our laws protect workers from intrusive and extreme surveillance from their employer? Mr Schmitke : The existing laws already do create protection. The various workplace surveillance acts right around the country at the state and territory level already provide quite significant protections for workers against unreasonable use of surveillance. There are penalties for employers that don't comply with those obligations, and I think those obligations are well known to most employers. The surveillance laws are well known to most employers, and I think they're entirely adequate already to provide those necessary protections. Senator SHELDON: Another example raised by the SDA is regarding Aldi, which has introduced an Amazon-style picking-efficiency surveillance tool. If you don't pick items quickly enough, you get formal warnings or you are eventually terminated. Of course, forcing everyone to pick faster and faster comes at the expense of safety. An SDA survey of affected Aldi workers found significant and escalating levels of musculoskeletal strain, extreme stress, mental exhaustion, extreme fatigue and more frequent workplace injuries. Workers that have been exceeding productivity benchmarks for years—these are actual workers that have been exceeding those benchmarks for years—were suddenly being told, as a result of this AI intervention, they weren't fast enough for the algorithm. The AI tool was creating issues—for example, where workers were being instructed to stack heavy items on the tops of pallets when placed on a truck and putting heavy loads on the back of a truck, which would make the truck unstable, which creates a major safety risk. Do you think that's a socially acceptable use of AI? Mr Schmitke : Again, I can't comment on whether that's socially acceptable. What I can say, again, is that we have a regulatory framework that's comprehensive and designed to protect against those circumstances. That particular situation would invoke three different regimes. One is the Fair Work regime, as I've identified previously, as to whether a fair process has been followed and adopted throughout that situation. The second is that we have a work health and safety framework, which is also very comprehensive. Clearly the examples that you've given involve potential hazards and risks. The approach that underpins the work health and safety framework now requires a comprehensive risk management approach. If there are safety issues that arise as a result of undertaking a risk assessment in that situation, the employer would be obliged to put in place changes to address those risks and minimise all the hazards that may arise. The third, of course, is the regulations that relate to the safety of heavy vehicles, loading capacities and things like that. Again, if circumstances arose where workers were potentially putting an operator of a heavy vehicle into breach of those laws, then they would be entitled to raise that as a concern. If they thought it was creating a risk or an imminent risk to somebody's safety, they'd be entitled to stop work until the matter had been resolved. Senator SHELDON: In the Aldi instance, and more generally, would you want to be micromanaged by cameras, wearable tech and an algorithm? Mr Schmitke : Again, I'm not an Aldi worker, so I can't answer that question. But I would say that I am aware of some of those examples from the SDA's [inaudible], which I think launched in parliament maybe 12 months ago or perhaps a bit longer now. I attended the launch of that event, as did our work health and safety director, because we're very interested in this issue. We did hear some stories about people who had been affected directly by these types of situations. We also heard in that particular event that a lot of these situations have subsequently been rectified. A lot of these situations were unintended consequences. There were good intentions behind what exactly was being introduced in a particular situation, and the issues that arose as a result of the implementation were (a) unintended and (b) subsequently addressed. As a result of that, significant improvements have been made where these types of technologies have been used. Conversely, we're also aware that there are many examples where these types of technologies make it easier and better for workers to understand what they need to do, when they need to do it and how it should be done. And it enables them to control their own workflow, enables them to have an idea and an understanding of what remains of their tasks for the rest of the day, and for some people that's actually a good thing. Senator SHELDON: So it would surprise you that, regardless of your statement just then, in the situation that I explained regarding the registered nurses—the use of AI where it hasn't been appropriately involving nurse practitioners and has been putting them at risk—and the SDA's example of the more generally retail use of AI for cashiers and the example of use of AI at Aldi, these large organisations are gaming the system? They're actually turning around and still using AI as an excuse to drive what they are describing as efficiencies but are in actual fact putting nurses' accreditation, health registration, clinical analysis and recommendations at risk because they don't have an opportunity to challenge them. In the case of the retail workers being oversighted by AI without their knowledge, and in the case of the warehouse workers not being able to have sufficient systems to actually challenge new AI systems that are put in place, that doesn't sound like a law that's working to me. Particularly with the monitoring and the wearing of wearables to monitor people, it sounds like it becomes more like a workplace that's got a digital ankle monitor. It's more like an AI prison that's being oversighted by AI governors, rather than the workforce having a say. It doesn't sound like the system is working to me. Does it to you, Mr Schmitke? It certainly doesn't sound like that to those workers. Mr Schmitke : Again, Senator, I can't speak for those individual workers, but what I would say is that none of the examples you've given me are examples that, in my view, represent circumstances where the existing laws aren't capable of dealing with those situations. If AI has been introduced and used to surveil workers in circumstances where they don't know about it, that would be a breach of the surveillance laws. That would also be deemed to be a fairly significant flaw, if any type of disciplinary action or termination action occurred, in the context of an unfair dismissal claim. There's clearly a procedural breakdown in those circumstances. I can't speak on the nurses example, because I'm unfamiliar with the specific regulation that applies to them in terms of their professional obligations. But once again I would say the general dispute resolutions under the Fair Work laws or under policies and procedures that pretty much all employers have in place provide for opportunities for workers to raise concerns about these types of technology and the circumstances in which they're being used. There are obviously obligations on employers to make sure that, if they're introducing new technology, people understand that technology and they understand how it's going to be used, what it's going to be used for and so on and so forth. If people introduce this type of technology and they rely on that technology without some sort of potential human oversight, that's going to be a key factor that would allow a challenge to the use of that technology moving forward. I can honestly see where you're going with all these examples, but I cannot genuinely think of a circumstance involving AI that can't be dealt with by the existing laws, simply because we need to be focusing on the conduct that might arise from the use of technology rather than the particular technology itself, and the conduct that arises in workplaces, such as that you've described, is all conduct, I believe, that's capable of being dealt with and rectified under the existing regime. Senator SHELDON: Just finally, if there's this breadth of problems and challenges with AI in the workplace, it says that the intent of the law needs to be considered—about what needs to happen to make sure the intent is being applied by what are major groups within industry. There's a serious question about the intent of the law and how it's functioning. Also, does that then require a review or revision of the law to make sure that employers are meeting those requirements? We're seeing a situation across the economy now where AI is being allocated in many workplaces, and it just appears that the existing laws aren't keeping up. It's a bit like when we went from horses to cars: of course, the laws needed to be equated for keeping on the road, regardless of the vehicles—particularly as a result of the vehicles changing. Mr Schmitke : Again, I'll just refer to my earlier comments. I would also note that the Australian government—I think as a result of its National AI Plan or statements that it released last year—indicated that it would be undertaking a review of the existing laws to identify whether or not they were capable of dealing with issues that might arise in relation to AI. I understand that review continues to be ongoing and is occurring on a whole-of-government basis. That's the second point I would make. The third point I would make is to make a confession to you here, and that is that I think the use of AI has created a lot of circumstances where employers and businesses are uncertain about the extent to which it triggers their existing legislative obligations under existing laws, and I feel there should be some better guidance given to business in this respect. You may be aware that the Australian government, through the NWRCC, has established an AI Employment and Workplaces Forum. That's being chaired by the Minister for Employment and Workplace Relations, Amanda Rishworth. I'm acting as part of that forum. I'm subject to all sorts of NDAs in terms of what happens at that forum, but what I can indicate to you is that the forum has agreed to develop some guidance for employers about how their existing regulatory obligations are triggered in the context of conduct arising from the use of AI. That's a process which is ongoing; it's only at its very early stages at the moment. So I think that would be very useful in helping business understand how AI related conduct should be treated under the existing law. The fourth point I'd conclude on is to say that the examples you've given today are not insignificant examples, but AI is being adopted across the entire Australian workforce and business community at a very rapid rate of knots. I've only heard of limited examples in particular sectors where it is problematic. I think there are a lot of other sectors where there aren't any problems. There are lots of workplaces where AI is being adopted at a very rapid rate, delivering lots of benefits in terms of improving productivity and benefits for employees. So I think we need to look at it in the context of that broader scale about the problem versus the question. Going back to what Mr Alexander said at the start in relation to regulation, review it and make sure it's adequate. If there are identified gaps, then let's have a look at it. But, at this point, we've not been able to see anything from a workplace perspective. CHAIR: Thank you. I understand there are no further questions, so we'll bring the session to a conclusion. Thank you so very much for giving up your time to us this afternoon, providing evidence and answering our questions. If there's any further work that you've committed to take on notice, please work with the secretariat to provide it within the next two weeks. Thank you very much..