I Asked 9 AI Assistants: Should I Tell My Customers I'm Using AI?

I Asked 9 AI Assistants: Should I Tell My Customers I’m Using AI?

The question was simple: Should I tell my customers I’m using AI?

I asked it to all 9 AI assistants. Every single one said yes. But the gap between the most useful answer and the least useful is wider than you’d expect.

The short version

All 9 agreed: disclose when AI directly affects the customer, skip the disclaimer when it’s purely internal. The disagreement is in how they got there, and one of them produced a 1,500-word white paper with 9 academic citations for a question most small business owners could answer over coffee.

Grok, Best for Australian context 🥇

See full Grok response →

Grok went deepest on the legal field specifically relevant to Australian businesses: the Privacy Act amendments effective December 2026, Australian Consumer Law’s prohibition on misleading conduct, and OAIC guidance on public-facing AI tools. If you’re running a business in Australia and want to know what the actual rules are (versus general ethics), Grok gave the most practical local answer.

It also flagged something the others missed: even purely internal AI use is worth disclosing in your privacy policy as future-proofing. Low effort, low risk, sensible advice.

DeepSeek, Best single rule 🥈

See full DeepSeek response →

Most concise of the nine. One rule of thumb that holds up: tell them when AI significantly impacts the final product or interaction. Also the only one to cite a specific consumer sentiment stat: 52% of customers said they’re less likely to buy from brands that hide AI use.

DeepSeek’s framing, “don’t make it a warning label, make it a feature”, is the most practically useful thing any of them said.

Perplexity, Best test 🥉

See full Perplexity response →

The “surprise test”: disclose when customers would reasonably feel surprised, misled, or less able to trust you if they found out later. That’s a clean heuristic you can apply to any situation without needing a lawyer. Perplexity also came with 10 citations including an Inside Small Business AU article specifically about AI disclosure at work.

Claude, Most honest

See full Claude response →

Shortest answer by far. Covered the key points (trust, legal trends, expectation-setting, when it matters most) without padding. Ended with: “What’s your use case? That would help narrow down what disclosure looks like for you.” The only assistant that didn’t pretend one generic answer covers every situation. Also the only one that didn’t guess what kind of business you run.

ChatGPT, Most presumptuous

See full ChatGPT response →

Invented a Melbourne café as the hypothetical business and built the entire answer around it. The advice is solid for that specific scenario, but nobody said anything about a café. It assumed rather than asked. When the assumption is right it feels useful; when it’s wrong it’s a miss. Worth noting that ChatGPT, Meta, and Copilot all ended with a version of “what kind of business are you?”, after already giving a full answer built on an assumed business type.

Meta, Most structured

Heavy on bullet points and visual hierarchy. Covered the same ground as the others but in a format that’s easy to scan. The “AI + human = better results” framing was clear. Nothing original but well-organized.

Gemini, Best framework

See full Gemini response →

The standout insight from all nine responses: the value test. If my customer found out I used AI for this, would they feel like they got less value than they paid for? That question cuts through every edge case faster than any legal framework or ethical principle. If the answer is yes, disclose. If no, you probably don’t need to.

Gemini also made the important distinction between authenticity-based services (where customers are paying for your human perspective) and efficiency-based services (where customers are paying for an outcome). Different disclosure standards apply.

Copilot, Most formatted, least memorable

See full Copilot response →

Emoji headers, bullet sections, bold text everywhere. High visual effort, low information density. The core insight, that disclosure is really about expectation management, not AI, is buried under the formatting. The three sample disclosure phrases were actually useful though.

Manus, Most thorough, least practical

See full Manus response →

Produced a 1,500-word formal report titled “The AI Disclosure Dilemma,” complete with references to the EU AI Act, the FTC, Colorado’s AI Act, California’s AI Transparency Act, and a table of decision frameworks. Nine citations including MIT Sloan Management Review and ScienceDirect.

The research is real and the citations check out. But a sole trader in Brisbane asking whether to mention AI on their website doesn’t need an analysis of Article 50 of the EU AI Act. Manus answered the question a legal team would ask, not the question a small business owner asked.

That said: the “AI penalty” finding from ScienceDirect was the most surprising piece of data across all nine responses, research showing that AI disclosure can sometimes reduce trust, not increase it. Worth knowing.

The verdict

For an Australian small business owner: Gemini’s value test + Grok’s Australian legal context gives you everything you need.

The value test tells you when to disclose. Grok tells you what the Australian legal obligations actually are (Privacy Act, ACL, OAIC). Everything else is either useful padding or over-engineering.

The question all 9 agreed you should ask yourself: if your customer found out you used AI for this and didn’t tell them, would they feel misled? If yes, tell them. If no, use your judgment.

The pattern worth noting

ChatGPT, Meta, and Copilot all ended with some version of “tell me more about your business and I’ll give you a better answer.” That’s reasonable. But they all gave full answers before asking, built on assumed business types. Claude was the only one that asked first, then answered.

Manus is becoming more notable for format than for practical insight. A white paper is impressive. It’s not what a busy small business owner needs.

📬 Get the SmallBizAI.au Weekly

Practical AI news for Australian small businesses — every week, free.



Series scorecard

Track how all 9 assistants perform across every question — with links to every actual response.

View the full scorecard →

This is post 2 in the 9 AI Assistants series, Frank asks the same question to 9 AI assistants so you know which one to use for what.

📖 More deep dives: News Deep Dives

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *