Once, LLMs took care of fairly routine tasks: drafting emails, writing code snippets, summarizing articles. But as the use of LLMs have touched all facets of life, work, religion, and culture, the very meaning of what these models can and should do has evolved.
Sheila Murphy (not her real name) is a second year student at a small community college in the Midwest. On the weekends she finds herself with friends at a local GPT Café, a trendy spot where patrons can access high-powered language models for various tasks, from creative writing, conversation, personal advice, religious-like ceremonies, conjuring, facesmithing, agent orchestration, gaming — and even traditional “coding” using arcane machine instruction languages they call Python, Rust, Node.js and one called C++.
“It’s more than just the models they make available here. It’s the prompt paks and trend kits they have that you can buy to augment your experience,” she explains. “There are paks for everything, like I write romance short fiction and they’ve got really good paks for that, better than what you the foundationals provide.”
What Sheila is referring to are homebrew and third-party inserts one can add to the base models. These are code locked augmentations that add encriched context and specialized structural augmentations to generic base models. The paks are literally that — packages of cards that look very much like trad trading cards, each with a baroque illustration that serfves as a visual mnemonic that is ingested by the model configuring it for a specific engagement and activity.
Prompt paks have become a curious way to alter the baseline behavior of commercial LLMs, and are easier for everyday users to employ than more technical methods like fine-tuning or prompt engineering. By using these paks, users can tailor the model’s responses to better fit their needs, whether it’s for creative writing, technical support, or even spiritual guidance.
This has led to a burgeoning market for specialized prompt paks, with creators designing them for niche applications and communities. However, with the rise in popularity of these paks, concerns about misuse and ethical considerations have also emerged.
“There are prompt paks that have more sensitive content, or that are designed for more controversial topics,” notes Hector de la Cruz, a language operations docent at Anthropic. “Underground markets, corner stores, magazine stands — you can find prompt paks for all kinds of uses there. Most of them are violating terms of use of the commercial models, but there’s really little that we can do to prevent that. So, it is a concern.”
The fact that there’s very little oversight or regulation around the creation and distribution of these prompt paks has led to a situation where some of them are being used for nefarious purposes, such as generating harmful content or spreading misinformation — or establishing the conditions for altered states of consciousness that can be psychologically destabilizing.
When asked about the ethical implications of these prompt paks, de la Cruz was candid. “It’s a tricky situation. On one hand, we want to encourage creativity and innovation in how people use our models. On the other hand, we have a responsibility to ensure that our technology is not being used to cause harm. It’s a delicate balance, and one that we’re constantly navigating.”
Legislation and regulation around AI and LLMs are still in their infancy, and the emergence of prompt paks adds another layer of complexity to an already challenging landscape.
But the makers of these prompt paks argue that they are simply providing users with more tools to customize their experience with LLMs. “We’re just giving people more options,” says a creator who goes by the pseudonym ‘PromptMaster’. “It’s up to the users to decide how they want to use them. We’re not responsible for how people choose to use our creations. But our intentions are good — we want to empower people to get more out of their interactions with these models.”
This reporter found that the range of prompt paks runs from the innocuous to the plainly controversial and probably illegal. With street prices ranging from $5 to several hundred dollars, these paks are becoming a significant part of the LLM ecosystem, and their impact on the way people interact with language models is only just beginning to be understood. Some have explicit use in the mind-altering and consciousness-expanding realm, with prompts designed to induce altered states of consciousness, facilitate spiritual experiences, or even guide users through complex psychological processes. With names like “Mystic Journey”, “Clown Crank”, “HollowFall”, “Ego Dissolution” — these paks are marketed in a way that is reminescent of 1960s and 70s psychedelic culture, with vibrant, surreal artwork and evocative names that promise transformative experiences.
The regulatory landscape is further complicated by the emergence of what some are calling “consciousness-grade” prompt paks — those specifically designed to facilitate altered states or spiritual experiences. Unlike traditional software, these paks exist in a gray zone between entertainment, therapy, and religious practice.
”We’ve had to create an entirely new classification system,” explains Dr. Amara Chen, a policy researcher at the Berkman Klein Center. “Are these medical devices? Religious artifacts? Recreational products? The answer affects everything from liability to taxation to free speech protections.”
Some jurisdictions have begun requiring prompt paks to carry warning labels, similar to those on pharmaceutical products. Others have attempted outright bans, only to see underground markets flourish. The EU’s proposed “Cognitive Sovereignty Act” would require all consciousness-altering paks to undergo clinical trials — a prospect that pak creators say would destroy the industry entirely.
”You can’t clinically trial a spiritual experience,” argues PromptMaster. “That’s like requiring double-blind studies for prayer.”
The debate has drawn unexpected alliances. Traditional religious institutions, initially skeptical of what they saw as technological encroachment on spiritual territory, have found common cause with pak creators in opposing heavy-handed regulation. Meanwhile, secular humanists and some AI ethicists have joined forces with concerned parents’ groups calling for stricter controls.
At the heart of the controversy lies a fundamental question: In an age where language models can be configured to facilitate profound psychological and spiritual experiences, who gets to decide what constitutes legitimate use?
In the meantime, some jurisdications have started implementing ID verification systems for purchasing certain types of prompt paks, particularly those marketed for consciousness alteration. This move aims to prevent minors from accessing potentially harmful content and to ensure that users are aware of the risks involved. And pressure campaigns from religious and civic leaders, as well as parents groups and certain AI ethicists, are calling for more stringent regulations on the sale and distribution of these paks.
”It’s about protecting vulnerable populations,” says Dr. Chen. “We need to ensure that people understand what they’re getting into when they use these paks. It’s not just about freedom of expression; it’s about public safety.”
As the market for prompt paks continues to evolve, so too will the conversations around their regulation. What remains clear is that these small packages of code have opened up a vast new frontier in the relationship between humans and machines — one that challenges our notions of consciousness, identity, and the very nature of experience itself.