Submission 183
PeopleForward AI
Submitter group: Companies.
An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.
Summary
Read the original submission for the full text.
Evidence from a two-person AI coaching and advisory business, which also asks government to support small AI advisory providers like itself. Argues adoption in small and family businesses is held back by not knowing where to start and by enterprise-scale governance frameworks; raises shadow AI, psychosocial factors, the Privacy Act small-business exemption and a quality standard for AI training providers.
Key points
Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.
Says that, in its client work, small and family business adoption is stalled more by not knowing where to start than by the cost of tools.
Across our client work, adoption is stalled far more often by not knowing where to start than by the cost of the tools themselves.
Argues that small businesses face a choice between full enterprise-grade AI certification and no formal governance framework, with no accessible proportionate option.
More broadly, we see a real gap between a small business's two realistic choices for AI governance: pursue a full enterprise-grade certification such as ISO/IEC 42001, which is a significant undertaking for a business our size, or have no formal framework at all.
Argues that unmanaged use of AI tools by staff and owners is a growing risk to copyright, data sovereignty and data security in small businesses.
Unmanaged use of AI tools by staff and business owners, sometimes called shadow AI, is a growing and largely unmanaged risk to copyright, data sovereignty and data security in small businesses that have nobody responsible for managing it.
Argues that the Privacy Act's $3 million turnover exemption lets very small businesses hold sensitive information and use AI tools on it without the obligations that apply to larger organisations.
The Privacy Act's three-million-dollar annual turnover exemption threshold for small businesses sits oddly against the reality of AI-era risk. A business well under that threshold, including a two-person business like ours. can hold sensitive client information and use AI tools on it, with none of the formal obligations that would apply to a larger organisation doing exactly the same thing.
Argues that the productivity benefit a business gets from AI depends on the skill, confidence, support and psychological state of the people adopting it.
How much a business benefits from AI depends heavily on the skill level and confidence of the person using it, how much support they have to learn it, and their own psychological state going in.
Says much current AI skills funding and curriculum targets schools and universities, leaving employed adults and small business owners with little accessible, non-technical training.
There is also a real gap in who current AI skills funding and curriculum reaches. Much of it is directed at schools and universities, which matters for the future workforce but leaves already-employed adults and small business owners, the people making adoption decisions right now, with little accessible, non-technical training built for their situation.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Address the psychosocial and change-management side of AI adoption (confidence, skill level, workload, recovery) in resilience and productivity guidance; any workforce AI adoption plan requirement should cover job design, workload, consultation and psychosocial safety.
Address the psychosocial and change-management side of AI adoption (confidence, skill level, workload and recovery) in resilience and productivity guidance, alongside technical capability. Any requirement for a workforce artificial intelligence adoption plan should cover job design, workload, consultation and psychosocial safety, not only the technical roll-out.
Develop a lightweight, proportionate AI governance and risk pathway for small and family businesses, scaled down from enterprise frameworks such as ISO/IEC 42001, rather than a choice between full certification and no framework.
Develop a genuinely lightweight, proportionate AI governance and risk pathway for small and family businesses, scaled down from enterprise frameworks such as ISO/IEC 42001, rather than leaving small businesses to choose between full certification and no framework at all.
Fund accessible, non-technical AI skills training for already-employed adults and small business owners, including grants or vouchers for independent advice and workforce capability building.
Fund and support accessible, non-technical AI skills training aimed at already-employed adults and small business owners, not only school- and university-based programs. Include grants or vouchers that can pay for independent advice and workforce capability building, not only software purchase.
Give small businesses clear, simple guidance on unmanaged ('shadow') AI tool use and its risks to copyright, data sovereignty and data security.
Give small businesses clear, simple guidance on unmanaged (‘shadow’) AI tool use and its risks to copyright, data sovereignty and data security, since these businesses have no information technology function to manage this themselves.
Review whether the Privacy Act small-business exemption threshold still fits, and raise visibility and support for the existing voluntary opt-in pathway.
Review whether the Privacy Act's small-business exemption threshold still fits a world where a very small business can hold sensitive information and use AI tools on it, and raise the visibility and support available for the existing voluntary opt-in pathway.
Provide a single plain guide joining privacy, consumer, discrimination, employment, copyright and cyber security obligations into one small-business pathway, kept current as laws change.
A small business does not experience AI regulation as separate legal subjects. A single everyday use can touch privacy, consumer, discrimination, employment, copyright and cyber security obligations all at once, and fragmented guidance for each one separately increases the chance an owner does nothing at all rather than works through six different sources. A single, plain guide that joins these obligations into one small-business-appropriate pathway, kept current as the law changes, would do more for genuine compliance than another standalone framework.
Recognise and support trusted, locally embedded AI coaches, trainers and small advisory businesses as part of the AI services sector, including through grants, procurement pathways and sovereign-capability support at that scale.
Recognise and support trusted, locally embedded AI coaches, trainers and small advisory businesses as a legitimate and valuable part of the AI services sector, including through accessible grants, procurement pathways and sovereign-capability support aimed at this scale of business, not only at large infrastructure and research providers.
Develop proportionate workplace AI training/advice quality standards covering trainer competence, conflicts or tool affiliations, evidence for teaching, and disclosure of material AI use in preparing materials, without excluding small or new providers.
A proportionate national quality and disclosure expectation for organisations offering this kind of workplace artificial intelligence training and advice, covering trainer competence, conflicts of interest or tool affiliations, evidence for what is taught, and disclosure of any material use of artificial intelligence in developing the training or advisory material itself, would raise the floor across the sector without shutting out small and new providers.
Consider how a small business or training provider without research capability could access, or contribute data to, narrow Australian domain-specific AI models.
We would encourage the Committee to consider how a small business or training provider without its own research capability could realistically get access to, or contribute data to, this kind of narrow Australian model,
Provide clearer small-business-appropriate procurement guidance and pathways for small AI-specialist businesses to compete for and deliver government work.
Clearer, small-business-appropriate procurement guidance, and pathways that allow small, artificial-intelligence-specialist businesses to compete for and deliver this kind of work themselves, would strengthen both sides of that relationship.
Model accessible government AI transparency: explain use, role, human review and supplier documentation, in a form small suppliers can meet and reuse.
Government is also in a position to model the standard it wants industry to meet: clearly explaining when artificial intelligence is used in a government service, what role it plays, how a person can ask for human review, and what its own suppliers are expected to document. Written so a small supplier can actually meet it, that standard becomes a template smaller businesses can reuse, rather than one more thing to interpret from scratch.
Offer recognition pathways and enablement support specifically for small, expert-led Australian AI services businesses, alongside support for larger infrastructure and research projects.
Grants, recognition pathways and enablement support aimed specifically at small, expert-led AI services businesses, not only at large infrastructure or research projects, would help retain this kind of domestic capability rather than see it wither for lack of a pathway to grow.
Provide a simple, low-effort way for very small businesses to identify everyday tools safe for client information, without interpreting a full enterprise standard.
This is not a training gap that can be closed with a single awareness campaign, it needs a genuinely simple, low-effort way for a very small business to know which everyday tools are safe to use with client information and which are not, without needing to read or understand a full enterprise standard to work it out.
Design retraining around changing tasks and transferable existing skills, rather than treating entire roles as obsolete.
Retraining design needs to start from that reality: what tasks are changing, and which of a person's existing skills carry forward, rather than treating the whole role as obsolete.
Give small businesses clear, simple guidance to avoid accidentally reproducing protected works with generative AI.
Beyond that specific risk, small businesses using generative AI tools in good faith need clear, simple guidance on how to avoid accidentally reproducing someone else's protected work, since very few have access to legal advice on this point before a problem arises.
Provide clearer, transparent licensing pathways for Australian creators' works used in AI training, enabling legitimate, straightforward use by small businesses.
Australian creators would also benefit from clearer, more transparent licensing pathways for their work being used in AI training, which would give small businesses a legitimate, straightforward way to use Australian creative and cultural material with confidence rather than uncertainty.
Provide practical, non-technical training for already-employed audiences in spotting AI-enabled fraud and reading digital content critically.
We also see, in client organisations, real exposure to misinformation, deepfakes and scams enabled by AI, and a genuine need for practical, non-technical training in how to spot AI-enabled fraud and read digital content more critically; again, training that is currently scarce for an already-employed, non-technical audience.
Provide accessible, non-technical guidance on securely configuring AI tools used by small businesses, alongside protection against sophisticated attacks targeting larger organisations.
Accessible, non-technical guidance on securely configuring the AI tools small businesses already use, not only on defending against sophisticated attacks aimed at larger targets, would close a real and currently under-served gap.
Explain AI-specific security risks simply, including instruction circumvention, connected-service data leaks and reliance on an external provider without fallback, so small businesses can act before problems occur.
A basic, non-technical explanation of risks specific to AI, such as a tool being tricked into ignoring its own instructions, information leaking out through a connected service, or a whole business depending on one external provider with no fallback, would help small businesses ask the right questions before a problem occurs, not only after.