The ethical red flag checklist for free AI tools

Free AI can be genuinely useful. But “free” is a pricing decision, not proof that a tool is safe, private or trustworthy.

AI has made it easier than ever to write, design, research and automate everyday work. That is exciting. It is also why people need to slow down before uploading files, handing over card details or trusting a tool simply because it looks polished.

The biggest risk is not that every free AI tool is a scam. Plenty are legitimate and useful. The risk is assuming that a free tool has the same privacy protections, security standards and accountability as a product built for professional use.

Before you sign up, upload, download or subscribe, use this checklist.

1. Ask what happens to anything you upload

A free AI tool may need a way to fund itself. Sometimes that means advertising. Sometimes it means paid upgrades. Sometimes it can mean using prompts, files or interactions to improve its products, subject to its terms.

Do not assume your uploads are private just because you have an account and a password. Read the tool’s data-use policy and look for clear answers to these questions:

• Can the company use your prompts, files or outputs to improve or train its models?
• Is there a setting to opt out, and is it switched on by default?
• How long is your information stored?
• Can you delete it?
• Does the company share data with third parties or subprocessors?
• Where is the data stored?

Policies vary widely, even within the same company. For example, OpenAI says personal ChatGPT workspaces have data sharing enabled by default, although users can opt out; its Business, Enterprise, Edu and API products do not use customer inputs and outputs for training by default. That is a useful reminder to check the exact plan and settings you are using, rather than relying on the product name alone. OpenAI data controls

For Australian organisations, the Office of the Australian Information Commissioner recommends, as best practice, that personal information — particularly sensitive information — not be entered into publicly available generative AI tools because of the privacy risks involved. OAIC guidance

A simple rule helps: if you would not comfortably post it in a public forum, do not paste it into an AI tool until you understand its terms.

2. Do not confuse “free” with “safe enough”

Free tools are often great for low-stakes tasks: brainstorming a holiday itinerary, improving a rough email or generating a first-pass list of ideas.

They are a much poorer fit for confidential material: customer records, staff information, unpublished financials, contracts, health details, passwords, source code or sensitive internal strategy.

The ethical question is not simply, “Can this tool do the job?” It is, “Do I have the right to put this information into someone else’s system?”

If the answer involves another person’s personal information, their consent and reasonable expectations matter. If you are using AI at work, use an approved tool and follow your organisation’s policy. If there is no policy, that is a reason to pause, not a reason to improvise.

3. Be honest about where AI has done the work

AI can help make work faster. It cannot replace judgement, experience or accountability.

If you use AI to help draft a customer-facing article, image, proposal or recommendation, the final work still needs a human owner. Check facts, add your own expertise and make sure the result genuinely says something you stand behind.

Being upfront does not mean announcing every spelling check or grammar suggestion. It means being transparent when AI has materially shaped the work, especially when a customer may reasonably assume they are receiving human expertise, original reporting or a personal recommendation.

Customers tend to notice when a brand has replaced its voice with generic, interchangeable text. The ethical use of AI is not about pretending the tool was never involved. It is about making sure the human contribution is still real: the judgement, the perspective, the fact-checking and the responsibility.

Australian privacy guidance also stresses transparency, including clearly identifying public-facing AI tools such as chatbots to customers. OAIC transparency guidance

4. Treat polished websites and ads with healthy suspicion

Scammers have always copied trusted brands. AI now makes it faster to create convincing websites, product images, fake reviews, emails and customer-service messages.

Scamwatch warns that criminals are using AI to create fake videos, clone voices, send personalised messages and build official-looking websites and documents. Scamwatch’s AI scam guidance

Before downloading a new AI tool:

• Go directly to the company’s official website rather than clicking a sponsored ad.
• Check the exact web address for odd spellings or extra words.
• Look for a real company identity, support contact and clear terms.
• Search the name alongside “review”, “scam” and “complaint”.
• Avoid downloading software from pop-ups, social posts or unfamiliar app stores.
• Never hand over remote access to your computer because a “support agent” asks.

The US Federal Trade Commission has warned that malicious ads for fake AI tools can send people to cloned websites and install malware. FTC consumer alert

5. Verify people, not just videos

A realistic voice note, video call or LinkedIn message is no longer proof that someone is who they say they are. A short public recording can be enough to create a convincing voice clone.

If someone contacts you urgently — even if they appear to be a colleague, friend, supplier or public figure — verify the request through another channel. Call a known number, start a new email thread or speak to them in person. Do not use the contact details supplied in the suspicious message.

Pressure is a red flag. So are requests for cryptocurrency, gift cards, passwords, one-time codes or immediate bank transfers.

6. Be wary of AI “experts” selling miracle outcomes

AI is a valuable skill to build, but it is not a 28-day shortcut to effortless income, instant business growth or a fully automated career.

Be cautious when an AI course, coach or community promises guaranteed results, a secret system, limited spots or a big income with little experience. These offers often rely on urgency and testimonials rather than clear teaching, demonstrated expertise and realistic expectations.

The FTC advises consumers to be sceptical of coaching programmes that promise guaranteed income, large returns or a “proven system”. FTC guidance on coaching scams

A credible education provider should be able to explain:

• Who teaches the course and what relevant experience they have.
• Exactly what you will learn.
• How much it costs in total.
• Whether there are recurring fees or upsells.
• How to cancel.
• What realistic outcomes look like.

AI capability is built through practice: learning how to ask better questions, checking outputs, understanding where tools fail and applying them to real work. Anyone selling certainty is selling the wrong thing.

7. Check for subscription traps before entering your card details

“Free trial”, “$1 access” and “limited-time AI toolkit” offers can become expensive recurring subscriptions if the ongoing cost is buried in the checkout flow.

The ACCC describes subscription traps as arrangements where a consumer is misled into an ongoing subscription or finds it difficult to cancel. It recommends that ongoing fees be clearly and prominently disclosed. ACCC guidance

Before paying, check:

• The total monthly or annual cost after any trial.
• The renewal date.
• Whether the subscription renews automatically.
• How cancellation works.
• Whether the provider has a real support channel.
• Whether you can download an invoice and manage billing yourself.

Set a calendar reminder before the trial ends and review your bank statement afterwards.

The checklist: pause before you proceed

Before using a free AI tool, ask:

1. What happens to my data?
2. Am I allowed to upload this information?
3. Can I opt out of model training or data sharing?
4. Who built this tool, and can I verify them?
5. Am I downloading it from the official source?
6. Is the output accurate, original and appropriate to publish?
7. Should I disclose the use of AI to customers or colleagues?
8. Is someone pressuring me to act quickly?
9. Are the promised results unrealistic?
10. What will this really cost after the “free” offer ends?

The best AI tools are not necessarily the most expensive ones. But the cheapest option can become costly when it exposes sensitive information, weakens trust or locks you into a subscription you did not mean to buy.

Use AI with curiosity. Just bring the same caution you would bring to any new supplier, app or financial decision.