How to Spot 6 Red Flags Showing a Message Was Written by ChatGPT

AI tools such as ChatGPT can be incredibly helpful, provided they are used thoughtfully. However, you might notice that an increasing number of texts across different situations are starting to feel somewhat… impersonal. Do you sometimes struggle to pinpoint why? Perhaps it’s that inviting message you received from a friend or an email from your supervisor.

Many individuals tend to over-rely on ChatGPT’s functionalities, using it excessively. Here are some straightforward indicators to help distinguish whether a piece of writing was created by a human or generated by AI.

1. Em-dashes. ChatGPT has a fondness for using em-dashes, making them a quick tip for spotting AI-produced content. Unless, of course, the writer intentionally uses them to enliven sentences— which is uncommon for casual writers. If a text is riddled with em-dashes used excessively and inappropriately, it’s likely AI-generated.

2. Overly elaborate phrasing. For example, if an invitation starts with, “This isn’t just a party to celebrate Staffan’s birthday; it’s also a gathering to welcome spring,” I might suspect AI assistance. Such constructions often seem overly polished or formulaic.

3. Exceptionally positive tone. If you’ve interacted with ChatGPT, you’ll notice it tends to avoid delivering critical or negative remarks, often balancing statements with positive ones—like “on the one hand… on the other hand…”. Attempts to extract a firm stance may result in the text sounding hesitant or diplomatic, reminiscent of a cautious spokesperson.

4. Lack of personal warmth. It’s challenging to create messages with genuine personal tone using AI. Even with prompts to do so, the responses often contain hyperbolic language and lack meaningful personal insights.

5. Repetition and over-summarization. AI-generated texts often repeat ideas or rephrase the same point multiple times. If a brief email ends with a reiteration or several sentences expressing a similar thought with odd analogies, it might be crafted by an AI rather than a human.

6. Perfect grammar and spelling. Text produced by AI typically lacks grammatical errors, especially if checked with tools. If the message is flawless in language, there could be an underlying automated process behind it, rather than a natural human touch.

Various services claim to identify AI-generated content, but I remain skeptical of their accuracy. I once inputted a five-year-old article to one of these tools and was told it was 94% AI-written, despite ChatGPT not existing at that time.