AI thinks I’m a serial killer: The ethics & challenges of using AI for horror writing

If you are reading this article, you’re curious about using AI to brainstorm ideas and get inspiration for your next horror story. You’re also open to AI playing a role in your creative work. However, if you’ve tried using AI tools like ChatGPT for horror writing, you may have noticed some unexpected limitations—certain themes, violent scenes, or explicit horror elements may trigger content moderation filters. So, how do you effectively use AI to enhance your horror writing while working around these restrictions? Let’s dive in.

My journey with AI tools

I began experimenting with AI tools toward the end of 2022 and started using ChatGPT a few months later. At first, I used it for fun, testing its limits with questions I already knew the answers to, a practice I enjoy as a QA tester at heart.

I soon realized ChatGPT could answer questions typically handled by Google searches or online/IRL writing communities. I find ChatGPT’s conversational style appealing and often struggle to stay active in online or in-person communities, so using this tool has been a valuable alternative.

However, my social circle doesn’t include any horror writers, and discussing my writing ideas with friends who enjoy horror has led to unexpected reactions—to put it bluntly, they were a bit freaked out. So, I turned to ChatGPT.

Ethical issues with AI in creative writing

Idea generation and brainstorming

AI can serve as a collaborative partner, offering prompts and suggestions that inspire writers. For instance, AI tools can suggest unique plot twists or character developments, aiding us in overcoming writer’s block.

Note by ChatGPT: AI models are trained on vast datasets, which may include existing plots and character archetypes. This means AI-generated suggestions could inadvertently resemble pre-existing works.

Two things to consider when brainstorming with ChatGPT:

• 👻 Specificity: Craft unique prompts that are detailed and tailored to your story’s context. Instead of a generic request like “Suggest a plot twist,” consider “Propose an unexpected friend betrayal in a secluded mountain village setting during a winter holiday.”

• 👨‍💻 Combining elements: Merge disparate ideas to form unique scenarios. For example, integrating folklore from different cultures can lead to original plot developments. Just be mindful when getting inspiration from other cultures.

Unethical uses of AI in creative writing (according to ChatGPT)

• 🗒️ Plagiarism and misrepresentation: Presenting AI-generated content as original work without proper attribution deceives readers and undermines the value of human creativity.

• 🎭 Deepfake narratives: Using AI to fabricate content that falsely represents real individuals, such as generating fictitious interviews or endorsements, raises significant ethical concerns.

  • Example: The documentary Roadrunner used AI to recreate Anthony Bourdain’s voice without clear disclosure, sparking controversy. NPR Article.

• 🧑 Unauthorized use of likenesses: Creating content that mimics the style or voice of existing authors without permission can infringe on intellectual property rights and mislead audiences.

• 😢 Emotion manipulation: AI’s ability to elicit emotional responses raises concerns about crafting narratives that manipulate readers unethically.

  • Author Kazuo Ishiguro has expressed concerns about AI’s capability to manipulate emotions, highlighting the need for ethical considerations in its application. Read more about his perspective here. 

Getting flagged

Getting my queries flagged by ChatGPT is disruptive. Instead of an inspiring session, I’m suddenly in a courtroom, defending my innocence to an AI convinced I have nefarious intentions.

I’m neither a murderer, a serial killer, nor a monster. Even as a horror writer, I’m sure Stephen King has written worse things. Human Centipede and Salò are not the kinds of topics I write about in my fiction. Yet, ChatGPT often flags my questions, which still surprises me.

Addressing ChatGPT’s explicit content flagging

I asked ChatGPT if adding context to our account “memories” might help ease the flagging when brainstorming about dark horror fictional topics.

ChatGPT’s response: “Setting your user profile to indicate that you’re a horror writer does not exempt you from ChatGPT’s content moderation policies.”

So, no, adding to your memories that you are a horror writer or that you write explicit fictional content won’t stop ChatGPT from flagging your queries and abruptly ignoring your brainstorming prompt.

The problem with content moderation filters

The University of Pennsylvania expands on how ChatGPT’s content moderation filter is, in my words, unhinged. I recommend reading the article for more insight: “

In one instance, given a prompt drawn from a summary of an episode of Game of Thrones, ChatGPT declined to produce the script and responded with a red warning: ‘This content may violate our usage policies.’”
 🔗 Read the full article

How to work within ChatGPT’s limitations without hindering creativity

I asked ChatGPT for its recommendations on how to navigate these restrictions, and here’s what it suggested:

• 💬 Modify prompt language: Focus on suspense and psychological elements rather than explicit violence or gore.

  • Example: Instead of requesting a graphic depiction, ask for a tense and eerie atmosphere.

• ✍ Iterative development: Break down scenes into smaller segments, allowing for a gradual buildup of tension without triggering moderation filters.

• 🗣️ Use metaphors and implications: Employ figurative language to suggest horror elements without direct descriptions that might be flagged.

• 👂 Seek human feedback: Collaborate with human editors or writing groups to refine content while adhering to content guidelines.

Final thoughts

Next, I plan to try out these tips myself. This article came from my curiosity about ChatGPT’s content filters and how to effectively use AI tools for brainstorming horror topics.

I believe AI tools like ChatGPT could be incredibly beneficial to horror writing, whether for research, brainstorming, or editing. However, it’s crucial to use them mindfully and ethically.

By crafting specific prompts, drawing inspiration from diverse elements, and using AI as a source of inspiration rather than direct content generation, we can preserve our voices and creativity.