Everyone’s using AI at work now. ChatGPT, Copilot, Gemini, they’re becoming as common as email. People are using them to draft documents, answer queries, and streamline workflows.
But here’s what almost nobody’s doing: creating proper policies around AI use.
And that’s a problem. A big one.
The Current State of AI in Australian Workplaces
I’m reading about organisations where:
-
- Staff are feeding confidential client data into ChatGPT without realising it’s being used to train AI models
-
- Managers are using AI to draft performance reviews without any verification of accuracy or fairness
-
- HR teams are applying US-based AI advice to Australian employment law situations (spoiler: our laws are completely different)
-
- No one’s checking for algorithmic bias in recruitment tools
-
- Businesses are using AI-generated contracts and policies without understanding whether they’re compliant with Australian law
This isn’t just inefficient or unprofessional. Under Australian privacy and employment law, it’s potentially unlawful.
When AI Goes Wrong: Real Consequences
Here’s what’s already happening: Commonwealth Bank recently made headlines when a staff member used AI and provided a customer with an incorrect phone number. It made the news because it shows how easily AI errors slip through without proper oversight.
And that’s just customer service, relatively low stakes.
Now imagine the risk when AI is being used for:
-
- Employment termination decisions
-
- Performance management processes
-
- Discrimination or harassment investigations
-
- Workplace injury assessments
-
- Modern Award interpretation
-
- Enterprise Agreement negotiations
The legal and financial consequences become far more serious. We’re talking, unfair dismissal claims, Fair Work penalties, discrimination cases, privacy breaches, all because someone trusted AI without proper human oversight.
Why “Just Being Careful” Isn’t Enough
I hear this a lot: “Oh, we tell our staff to be careful with AI.”
That’s not a policy. That’s a hope.
Without clear boundaries and accountability frameworks, “being careful” means different things to different people. One person thinks it’s fine to use AI for drafting termination letters. Another thinks it’s fine to feed performance data into ChatGPT. Someone else assumes AI-generated legal advice is accurate.
They’re all wrong. But without a policy, how would they know?
What Australian Businesses Actually Need
Before implementing ANY AI tool in your workplace, you need:
1. A Comprehensive AI Policy
This sets clear boundaries on:
-
- Which AI tools are approved for use and why
-
- What types of information can and cannot be fed into AI systems
-
- Which tasks require human oversight before AI can be used
-
- Who’s accountable when AI gets it wrong
-
- How to handle confidential or sensitive information
2. Mandatory Human Oversight
And not just any human, someone with deep knowledge of Australian employment law and privacy law, not just general HR knowledge.
Because it depends on what prompts you use and not all AI will know:
-
- The difference between the Restaurant Award and SCHADS Award
-
- How procedural fairness applies in your specific situation
-
- Whether your enterprise agreement overrides standard provisions
-
- When you need to consult before making changes
-
- How cultural safety protocols should be applied
3. Australian-Specific Compliance Checks
ChatGPT was trained primarily on US data. When you ask it about employment law, it defaults to American assumptions unless you’re very specific.
But even then, it gets things wrong. I’ve seen AI confidently cite Fair Work Act provisions that don’t exist, misinterpret Modern Award classifications, and give advice that would expose businesses to significant penalties.
You need someone who actually knows Australian employment law to verify every AI output before it’s implemented.
4. Cultural Safety Protocols
This is particularly critical for organisations working with Australian First Nations communities or diverse cultural groups.
AI doesn’t understand:
-
- Cultural communication preferences
-
- The importance of relationship-building in certain contexts
-
- When “following the procedure” will damage trust irreparably
-
- How to adapt frameworks to be culturally appropriate
These require human judgement, cultural competence, and deep contextual understanding.
5. Risk Assessment Frameworks
Not all AI use carries the same risk. Using AI to schedule meetings? Low risk. Using AI to draft a termination letter? High risk.
You need a framework that helps staff assess risk and know when they need specialist advice before proceeding.
The Gap I’m Witnessing
Here’s the pattern I’m noticing across Australian businesses:
They think having an HRIS platform, subscription systems or using AI tools means they’re covered. They assume the technology will keep them compliant.
But technology handles transactions. It can’t replace human judgement about complex employment law, cultural nuance, or ethical decision-making.
Having systems doesn’t mean you understand Modern Award classifications.
AI-generated processes don’t mean you’ve applied procedural fairness correctly.
Automated compliance reminders don’t mean you know what you’re actually required to do.
And this is exactly when businesses get into expensive trouble.
The Real Cost of Getting This Wrong
Let me be blunt about what’s at stake:
-
- Fair Work penalties up to $99,000 per contravention and higher
-
- Unfair dismissal claims that cost $15,000-$50,000+ to settle
-
- Discrimination cases with uncapped damages
-
- Privacy breaches under Australian Privacy Principles
-
- Reputational damage when AI mistakes become public
-
- Loss of trust with employees, clients, or community partners
The cost of getting this wrong is significantly higher than the cost of getting it right.
What You Should Do Now
If your organisation is using AI without a proper policy framework, you need to address this urgently.
Here’s what that looks like practically:
Immediate steps:
-
- Audit current AI use across your organisation (you might be surprised what people are using)
-
- Identify high-risk applications (anything involving employee data, legal decisions, or confidential information)
-
- Implement temporary restrictions while you develop a proper policy
Medium-term:
-
- Develop a comprehensive AI policy tailored to your organisation
-
- Train staff on boundaries and requirements
-
- Establish accountability structures
-
- Create risk assessment frameworks
Ongoing:
-
- Regular reviews as AI technology evolves
-
- Incident reporting when AI produces problematic outputs
-
- Updates to policy as Australian law changes
Why I’m Talking About This
After 30+ years working in Australian businesses and 15+ years working with deep knowledge of Australian employment law, I’ve developed an AI policy framework that helps organisations use AI safely and legally.
I created it because I saw the gap: businesses adopting AI rapidly without understanding the legal and ethical implications.
And yes, I use AI myself in my practice. Every single day.
But here’s the critical difference: every AI output gets reviewed through the lens of 15 + years of Australian employment law knowledge. I verify facts. I check Australian-specific compliance. I ensure cultural appropriateness. I apply human judgement to contextualise the output.
That’s the difference between using AI wisely and using it recklessly.
The Bottom Line
AI is here to stay. It’s genuinely useful for many tasks. I’m not anti-technology.
But we need to use it responsibly, with proper policies, clear boundaries, and genuine human expertise providing oversight.
The businesses that get this right will use AI to work smarter while avoiding the costly mistakes that come from blind trust in technology.
The ones that don’t will be the case studies we read about in employment law updates.
Here’s My Question for You
Is your organisation using AI without a proper policy framework?
Are you confident your team knows the boundaries?
Can you verify that AI outputs are being checked by someone with deep knowledge of Australian privacy and employment law?
If the answer is “not sure” or “probably not,” let’s talk.
I’ve developed an AI policy framework specifically for Australian businesses, covering:
-
- Appropriate use boundaries
-
- Mandatory oversight requirements
-
- Australian-specific compliance checks
-
- Cultural safety protocols
-
- Risk assessment frameworks
-
- Accountability structures
Because the cost of getting this wrong is a lot higher than the cost of getting it right.
Contact me to book a FREE-15 minute discovery call