AI washing is when a vendor labels their product “AI-powered” without the technology to back it up. It’s one of the fastest-growing forms of misleading marketing in the tech industry right now, and Australian businesses are squarely in the crosshairs. The ACCC has made AI washing an explicit enforcement priority for 2026-27, and the legal consequences for vendors are real.

If you are evaluating software for your business, this guide will help you separate genuine AI capability from clever marketing language.

---

What Is AI Washing?

AI washing is the practice of vendors overstating, exaggerating, or misrepresenting the AI capabilities of their products or services to make them appear more sophisticated, valuable, or modern than they actually are. In plain terms, it is when a product is marketed as “AI-powered” but the underlying technology is basic automation, scripted logic, or rule-based filtering.

The term mirrors “greenwashing,” which refers to companies falsely claiming environmental credentials. Just as greenwashing exploits consumer trust in sustainability, AI washing exploits the current hype around artificial intelligence.

This is not a fringe problem. AI washing is widespread across cybersecurity, project management, HR tech, and productivity software. When evaluating vendors for our clients at Otto IT, we encounter AI washing claims regularly, and the gap between marketing language and technical reality is often significant.

Why AI Washing Is a Problem for Your Business

When you pay for AI that does not exist, or AI that does not perform as described, you are not just wasting money. You are making strategic decisions based on capabilities your tools do not actually have.

For professional services firms, this creates real risk. A cybersecurity tool marketed as AI-driven threat detection that is actually just signature-based antivirus will leave you with a false sense of protection. A project management tool claiming “AI insights” that delivers nothing beyond basic reporting will fail to deliver the productivity gains you planned around.

---

Is AI Washing Illegal in Australia?

Yes, AI washing is already illegal in Australia under existing law, even without dedicated AI legislation. The Australian Consumer Law prohibits misleading and deceptive conduct in trade and commerce. A vendor who falsely claims their product uses artificial intelligence when it does not is engaging in misleading conduct and may face enforcement action.

The ACCC’s 2026-27 Compliance and Enforcement Policy has made AI washing an explicit priority alongside dark patterns and subscription traps. This is significant. It signals that the ACCC intends to actively investigate and act on misleading AI claims, not simply wait for complaints.

In October 2025, the ACCC commenced legal proceedings against Microsoft, alleging the company concealed the existence of a cheaper Microsoft 365 plan that did not include Copilot AI features. The case is a clear signal that major vendors are not exempt from scrutiny.

Law firm Minter Ellison has also warned that companies overstating AI capabilities face legal risks not just under consumer protection laws but also under directors’ duties, as executives may be personally exposed if AI misrepresentation materially affects business decisions or investor value.

What This Means for Buyers

The legal risk sits primarily with vendors who make false claims. But buyers carry a different kind of risk: spending on tools that underperform, then having to explain to the board or partners why the technology investment did not deliver.

Understanding how to identify AI washing before you sign protects your business commercially and protects your reputation as a decision-maker.

---

What Are the Most Common AI Washing Tactics?

The most common AI washing tactics involve using technical-sounding language to obscure what is actually quite simple automation. Understanding these patterns makes them easier to spot in vendor demos and sales conversations.

“AI-powered” automation that is just rule-based logic. Many tools describe simple if-then automation as AI. If a system routes support tickets based on keywords, that is not machine learning. That is rules. The distinction matters because rule-based systems cannot learn, adapt, or improve without manual reconfiguration.

“Machine learning” used as a buzzword with no implementation. Some vendors claim their product uses machine learning without being able to describe what the model does, what data it trains on, or how it improves over time. Machine learning is a specific methodology. If a vendor cannot explain how their model is trained, there is a reasonable chance it is not real.

Inflated AI accuracy statistics. Claims like “95% accuracy” or “10x faster than manual review” are common and often unverifiable. Otto IT recommends asking for the methodology behind any accuracy claim. What dataset was used? What were the test conditions? Were the tests conducted independently?

AI features that exist in demo but not in practice. Some vendors showcase AI capabilities in sales demos that are not available in standard deployments, or that rely on configurations not achievable in most business environments.

Chatbots marketed as AI assistants that are scripted decision trees. There is a material difference between a large language model that can reason through an open-ended question and a scripted chatbot that matches keywords to pre-written responses. Both can be useful, but they are not the same thing. If a vendor calls their scripted bot an “AI assistant,” that is an AI washing claim.

---

Real-World Examples of AI Washing

These examples reflect patterns Otto IT has observed when evaluating vendor claims on behalf of professional services clients. They are based on category-level patterns, not individual named vendors.

Cybersecurity tools claiming AI-driven threat detection. Several endpoint security vendors market their products as using AI or machine learning for threat detection. In practice, many of these tools rely primarily on signature-based detection, which matches known threats against a database of patterns. Signature-based detection is a long-established technique with no machine learning involved. It cannot detect genuinely novel threats that have not appeared in its database. When evaluating vendors for our clients, Otto IT asks specifically whether the detection engine adapts without signature updates. That question quickly separates real ML from marketing.

Project management tools claiming AI insights. “AI-powered project insights” often means a dashboard that shows you which tasks are overdue or which projects are trending behind. These are basic reporting features. They involve data aggregation and visualisation, not pattern recognition, prediction, or natural language understanding. When vendors cannot explain how the AI recommendation was generated, that is a signal worth noting.

HR platforms claiming AI candidate matching. Some recruitment tools claim AI-driven candidate shortlisting. In some cases, this is genuine: the system has been trained on historical hiring data and generates probabilistic rankings. In other cases, it is a keyword filter that excludes candidates who do not use certain terms. Asking how the matching score is calculated and what it is trained on will quickly reveal which category you are dealing with.

---

How Do You Spot AI Washing Before You Buy?

The most effective way to spot AI washing is to ask specific, technical questions that require concrete answers. Vague marketing claims fall apart quickly when a vendor cannot answer basic questions about their own technology.

When evaluating vendors for our clients, Otto IT uses a structured set of questions to surface genuine AI capability. Here are the questions worth asking:

Questions to ask about the technology:

• What type of AI or machine learning model does this feature use?
• What data does the model train on, and how frequently is it retrained?
• Can the model improve its outputs without manual configuration by your team?
• What happens to the AI feature if the training data is removed or restricted?
• Has the AI capability been independently tested or certified?

Questions to ask about accuracy and performance:

• What is the documented accuracy rate, and how was it measured?
• What were the conditions of any accuracy testing?
• Can you provide a trial environment where we can test the AI claims against our own data?

Practical checks during product evaluation:

• Request a technical explainer document, not a marketing one-pager
• Ask to speak with a technical pre-sales engineer rather than a sales representative
• Ask whether the AI feature is available on the base tier or only on premium plans
• Check whether the vendor lists the AI as a separate, identifiable component in their product documentation

If a vendor cannot answer these questions clearly and specifically, that is meaningful information. Legitimate AI vendors are generally able and willing to explain how their technology works.

---

What Does Genuine AI Look Like in Business Software?

Real AI in business software typically shares a few characteristics that distinguish it from rule-based automation.

Genuine AI systems can handle inputs they have not seen before. A machine learning model trained on support tickets can categorise a new type of ticket it was not explicitly programmed to handle. A rule-based system cannot.

Genuine AI systems improve over time with more data. If a vendor cannot explain the feedback loop by which their system gets smarter, it probably does not have one.

Genuine AI systems have explainable outputs. This does not mean the algorithm is always interpretable, but vendors should be able to describe what factors the model considers and what it is optimised for.

Legitimate AI vendors are also typically transparent about limitations. They will tell you what their model is not good at, what edge cases cause problems, and what human oversight is still needed. If a vendor’s AI claims are absolute and without caveats, that itself is a signal worth examining.

---

What Should Australian Businesses Do Right Now?

Otto IT recommends three immediate actions for any professional services business that is currently evaluating or using AI-marketed software.

Audit your existing AI tools. If you are paying for a product marketed as AI-powered, ask your vendor to provide technical documentation that explains the AI component. If the documentation does not exist or is vague, you are likely paying for something that does not match the marketing.

Update your vendor evaluation process. Add a mandatory section to your procurement checklist that requires vendors to answer specific questions about their AI claims before a purchase decision. This protects you commercially and legally if a vendor later misrepresents their capabilities.

Stay across ACCC enforcement activity. The ACCC has signalled it will pursue AI washing cases actively. If you discover you have purchased software on the basis of misleading AI claims, you may have recourse under Australian Consumer Law. Keeping records of vendor claims made during the sales process is good practice.

For a full review of your technology stack and vendor agreements, speak with the Otto IT team. We evaluate vendor claims independently and help professional services firms make technology decisions that deliver real-world outcomes.

If you are currently in a procurement process and want a second opinion on vendor AI claims, book a consultation with our team.

For broader questions about technology strategy for your business, visit our IT services page at ottoit.com.au or contact us directly.

---

Frequently Asked Questions

Is AI washing illegal in Australia?

Yes. AI washing that involves false or misleading claims about a product’s AI capabilities is prohibited under the Australian Consumer Law, which bans misleading and deceptive conduct in trade and commerce. The ACCC has listed AI washing as a 2026-27 enforcement priority, meaning active investigation and legal action are likely. Businesses that purchase software based on false AI claims may have legal recourse.

How do I know if a product genuinely uses AI?

Ask the vendor to explain what type of AI or machine learning model underpins the feature, what data it trains on, and how it improves over time. Legitimate AI vendors can answer these questions with specifics. If the answers are vague, or the vendor redirects to marketing material rather than technical documentation, treat that as a warning sign.

What is the difference between AI and rule-based automation?

Rule-based automation follows a fixed set of instructions written by a human. It can only handle situations it was explicitly programmed for. AI and machine learning systems, by contrast, are trained on data and can handle new situations they have not seen before. They can also improve over time without being manually reprogrammed. Many tools marketed as AI are actually rule-based systems.

Can I take action if I have already bought a product that overstated its AI capabilities?

Potentially, yes. If a vendor made specific claims about AI capabilities during the sales process and the product does not deliver on those claims, you may have grounds for a complaint under the Australian Consumer Law. Document any claims made in writing, including emails, demo recordings, and sales materials. Speaking with a lawyer who specialises in consumer or commercial law is advisable before taking action.

Why is AI washing such a big issue right now?

AI washing has accelerated because “AI-powered” has become a premium selling point that commands higher prices and faster purchase decisions. Vendors have strong commercial incentives to label their products as AI even when the underlying technology does not qualify. At the same time, most buyers lack the technical knowledge to verify those claims, which makes the market fertile ground for misrepresentation. The ACCC’s decision to make AI washing an enforcement priority reflects how widespread the problem has become.

---

The Bottom Line on AI Washing

AI washing is not a minor marketing quirk. It is a form of misleading conduct that is illegal under Australian Consumer Law and increasingly in the ACCC’s enforcement crosshairs. For professional services firms, buying software based on inflated AI claims creates real financial and operational risk.

The fix is straightforward: ask better questions before you buy. Require vendors to explain their AI in specific, verifiable terms. Build those questions into your procurement process. And if you are not sure what to ask or how to evaluate the answers, Otto IT can help.

We work with professional services firms across Australia to evaluate technology vendors independently, cut through marketing language, and build IT environments that actually perform. If your business is navigating vendor AI claims right now, book a conversation with our team or contact us here.

AI washing is the vendor’s problem legally. It becomes your problem commercially if you do not catch it first.