Most companies now have, or are drafting, rules for using AI at work. What happens at employees' desks is often something else. Someone is pasting a sales report into a free chatbot to tidy it up. Someone else is asking a personal AI account to rewrite a customer email. Nobody told IT.
This is Shadow AI: employees using generative AI tools the company has not approved. The original Thai article cites a claim that such use has grown about fourfold, though it gives no source for that number.
Why it is a real risk
The danger is a data-flow problem, not an employee-laziness problem. People are trying to work faster with what is available.
- Trade secret and personal data leakage. When staff paste sales figures, product plans or customer data into a public AI tool, that information leaves the company's control. Depending on the tool's terms, it may be used for further training.
- Copyright and compliance exposure. Content or code produced by outside tools can carry overlapping copyright. If IT does not know a piece of work is AI-generated, the company may face legal questions later.
Why banning does not work
If people are reaching for AI because it makes them faster, a ban mostly pushes the behavior further out of sight. The more workable approach is to give people a safe, approved option and clear rules:
- Provide a company-approved AI tool so people do not have to improvise
- Define what data must never be pasted into any external tool
- Train staff on why, not just what
- Review contracts and settings of approved tools for data handling
The creator's version of the same problem
Cheap generation has another side effect: content pollution. The internet is filling with unscreened, emotionally flat and sometimes inaccurate synthetic content. For creators and brands, chasing views by publishing faster no longer works.
The original article proposes two answers:
- Authority. Audiences are tiring of shallow, fast content. They want in-depth analysis from real people with real identities, who cite and verify their sources.
- Owned media. Relying only on social platforms is increasingly risky. A personal website and a direct audience database protect you from algorithm shifts.
A note on evidence
This is a short, opinion-driven piece. It offers one unsourced statistic and no named studies or case examples, so treat the fourfold figure as a prompt for your own audit, not as established data. The framework, though, is sound: separate the risk to organizations (data and compliance) from the pressure on creators (credibility and ownership).
Questions worth asking inside your organization
- Do we have an AI usage policy that people actually know about?
- Which AI tools are employees using today, and do we know?
- How do we tell human-written work from AI-generated work when it matters?
My take
Convenience and speed are not always the best answer. Enterprises need oversight and ethical guidelines that make the safe path the easy path. Creators survive by putting accuracy, consistency and credibility first, the things people can rely on in real life, instead of racing bots at volume.
Written by Ko John. Based on the Thai original on kojohn.com.
Original Thai analysis available at /articles/shadow-ai-enterprise-creator-trust
