ChatGPT is finally pushing back on requests to mimic specific

OpenAI seems to be tightening the reins on how ChatGPT handles style mimicry. If you've tried to tell the model to "write exactly like [Famous Author]" recently, you might have noticed it becoming more evasive or giving you a generic response instead of a spot-on imitation. It's a clear shift toward protecting intellectual property and avoiding the "plagiarism-by-proxy" trap that happens when LLMs are used to clone a living writer's unique voice for commercial gain.

For those of us obsessed with prompt engineering, this is actually a blessing in disguise. Relying on the model's internal knowledge of a person's style is lazy prompting anyway—it usually results in a caricature of the author (think: too many adjectives for Hemingway or overly dramatic prose for someone else) rather than a nuanced adaptation. To get around these blocks and actually improve your AI workflow, you need to move from "Who" to "How."

Instead of naming a person, you have to describe the linguistic DNA of the style you want. Here is a practical tutorial on how to build a style profile from scratch so you aren't dependent on the model's "celebrity" presets.

How to build a manual style profile

1. Analyze the target text: Take a sample of the writing you admire. Identify the average sentence length, the ratio of active to passive voice, and the specific vocabulary level (academic, visceral, minimalist).
2. Define the constraints: Create a set of rules. For example, "Avoid adverbs," "Use short, punchy sentences," or "Start paragraphs with a provocative question."
3. Feed the pattern, not the name: Use a few-shot prompting technique. Provide 3-5 examples of the style you want, followed by the task.

Here is how you should structure your prompt now to avoid the "blocking" behavior:

Act as a professional editor. I want you to write a piece based on the following stylistic guidelines:
- Sentence Structure: Vary between very short (3-5 words) and medium (10-15 words). No long, winding complex sentences.
- Tone: Cynical but informative. 
- Vocabulary: Use concrete nouns and avoid abstract corporate jargon.
- Rhythm: Use a staccato pace to create urgency.

Here are three examples of this style:
[Insert Example 1]
[Insert Example 2]
[Insert Example 3]

Now, write a 300-word analysis of the current LLM agent market using this exact framework.

This approach turns the process into a deep dive into linguistics rather than a request for a copy-paste clone. By defining the parameters—like cadence, lexicon, and tone—you get a much more consistent output that doesn't trigger the "I cannot mimic this specific person" guardrails. It forces the LLM to actually process the structure of the language rather than just pulling from a training set of known authors. This is the only way to ensure your content doesn't feel like a generic AI hallucination of a famous writer.