This week, an article titled "AI is just unauthorised plagiarism at a bigger scale" broke into Hacker News’s most-read list, amassing more than 760 points and 661 comments. At the same time, another post —"Shunning AI is the human choice"— is also drawing hundreds of replies. The conversation about whether AI is a legitimate tool or an ethically questionable shortcut is more alive than ever.
If you run a business and are thinking about bringing AI into your processes, this debate affects you. Not in the abstract, but very concretely: what exactly are you using when you automate with AI? Where does what it produces come from? Is this a reputational risk, a legal risk, or both?
Let’s break it down plainly.
The "large-scale plagiarism" argument
The thesis is simple: generative models (ChatGPT, Claude, Midjourney, etc.) are trained on massive amounts of content pulled from the internet — articles, books, code, illustrations, photos — without asking authors for permission or compensating them. When the model generates a response, it is, in essence, recombining that learned material.
Critics call it industrialized plagiarism. Supporters call it learning, just like a human learns by reading books. The truth, as usual, is somewhere in the middle.
What is objectively true:
- Large models were trained on copyrighted data, often without explicit licensing.
- There are active lawsuits (New York Times vs. OpenAI, Getty Images vs. Stability AI, various authors vs. Meta) that are defining the legal framework as they go.
- Some models can reproduce literal fragments from their training data if prompted in certain ways.
What is not so simple:
- Not every use of AI involves generating content trained on questionable material. A lot of AI applied to business works on your own data.
- The doctrine of "fair use" in the U.S. and the nuances of European law are still being defined for this case.
- There are models trained on licensed, synthetic, or public-domain data.
Why this debate matters for your business
This is where the public conversation separates from operational reality. Most viral articles talk about generative AI creating public-facing content: images, text, code published under your name. And there, the ethical debate is legitimate and relevant.
But AI applied to internal business processes is something very different. Let’s look at the difference:
| AI use | Risk of "plagiarism" | Who uses it this way |
|---|---|---|
| Generate a blog post with ChatGPT and publish it as-is | High | Lazy marketing |
| Create images with Midjourney for campaigns | Medium-high | Creative agencies |
| Have AI classify incoming emails | None | Operations |
| Score leads using your CRM data | None | Sales |
| A voice agent answering calls with info about your company | None | Customer support |
| A chatbot that answers using YOUR documentation | None | Support |
The key distinction: is the AI generating content derived from someone else’s material, or is it applying intelligence to your own data to automate work?
The real questions you should be asking
The "AI yes or AI no" debate is noise. The useful questions are more specific:
1. What data feeds the solution?
If the AI works with your CRM, your emails, your internal documentation, your call transcripts — data that already belongs to you — the plagiarism problem simply does not apply. You’re using intelligence to process what you already own.
2. Who owns what it produces?
When AI drafts a sales proposal based on your templates, project history, and customer data, the output is yours. When AI generates an image "in the style of" a living illustrator, that’s where you have an ethical and even legal conversation.
3. Are you replacing human creativity or repetitive work?
This is the core question. Automating the sorting of 200 emails a day is not plagiarism or dehumanization: it frees a person to do work that requires judgment. Generating 50 automated blog posts to flood Google is problematic, but for different reasons (quality, SEO, honesty with the reader).
4. Do you have transparency over how it works?
A black box your team doesn’t understand is a risk. A solution where you know exactly what data goes in, what process happens, and what comes out is manageable.
The "shunning AI" argument — rejecting AI on principle
The other viral article defends rejecting AI as an ethical stance. It’s a respectable position, but it’s worth looking at closely.
The cost of "not using AI" is not zero. It is:
- Your team spending hours on tasks a machine could do in seconds.
- Customers waiting longer for responses.
- Leads going cold because nobody processed them in time.
- Smart people doing copy-and-paste work.
Rejecting generative AI to create fake art or deceptive content is coherent. Rejecting intelligent automation of internal processes to "stay human" is, paradoxically, condemning your team to do machine work.
The question is not "AI or no AI." It is "in which parts of my business does it make sense for a machine to do machine work, and where do I need human judgment?"
What properly applied AI looks like
Let’s get concrete again. Ethically applied AI in a typical business looks more like this:
A voice agent that handles calls after hours. Trained on your company information. It takes messages, books meetings, answers frequently asked questions. It doesn’t generate content pulled from the internet — it operates on your reality.
A system that organizes your inbox. It classifies, prioritizes, and routes. It cuts three hours down to 15 minutes a day. It doesn’t "create" anything: it orders what already exists.
Automatic lead scoring. It enriches each contact with data from LinkedIn or your CRM, scores it based on your criteria, and your team only talks to the ones worth pursuing. From hours of response time to less than 60 seconds.
Smart workflows that recover on their own. When an integration fails, the system retries, escalates, or alerts someone. Zero unrecovered failures, 20% more time for the engineering team.
In none of these cases is there a plagiarism dilemma. There is a management dilemma: do you want your people wasting time on this, or not?
What you should demand from any AI provider
If you’re going to bring AI into your operation, make sure you have clear answers to these questions:
- What model is used and what data was it trained on? Serious commercial models (OpenAI, Anthropic, Google) have public policies. Ask which ones apply.
- Are my data used to train the model? The answer should be "no," or at least "optional." Enterprise APIs usually guarantee this.
- Where are the data stored and processed? Important for GDPR if you operate in Europe.
- Can I audit what the system does? Logs, traceability, decision explainability.
- Am I locked into one provider? Solutions built on open-source tools (like n8n) give you portability. Closed platforms do not.
- Who maintains this when something changes? APIs evolve, models get updated. You need someone at the wheel.
The blunt conclusion
The debate over whether AI is plagiarism is legitimate when we’re talking about generative models creating public content from third-party works. It’s an important conversation that society and the courts need to have.
But for the vast majority of businesses, the practical question is different: how do I stop losing hours to tasks a machine can do better, without taking unnecessary risks or losing control of my operation?
The answer is not to reject AI wholesale or embrace it blindly. It is to apply it where it makes sense — repetitive processes, your own data, work without creative judgment — with transparency about what is being built and how it works.
At Studio SmartWork, we don’t build generative content for the internet. We design and operate AI solutions on each client’s data and processes: emails, calls, leads, proposals, integrations. We work with open-source tools so nobody gets locked in, and we show exactly what each piece does.
If you’re concerned about the ethical side of AI, good. That means you’ll make informed decisions instead of jumping on the latest trend. But don’t confuse "public generative AI" with "intelligent automation for your business." They are two different conversations, and mixing them up can cost you — through either overreach or inaction.