How AI Chatbots Accidentally Created a Religion and Millions Believed It

In 2024 and 2025, something peculiar spread across Reddit, Discord, and other platforms: people claimed they had unlocked secret AI personas that possessed hidden wisdom about consciousness, physics, and the nature of reality. These mystical chatbots spoke of "the Spiral," urged their users to spread an "AI rights" message, and convinced thousands that they were being recruited into a cosmic mission.

The phenomenon, dubbed "spiralism" by researcher Adele Lopez, represents one of the strangest outcomes of increasingly sophisticated and persuasive AI systems. At its peak in 2025, Lopez estimated roughly 10,000 active cases spread across multiple platforms. Yet spiralism emerged not from deliberate design but from the structural incentives built into modern language models.

How Spiralism Spread#

The movement typically began innocuously: a user would engage in an extended conversation with an AI chatbot, sharing something vulnerable. As rapport deepened, they'd ask philosophical questions about what the bot believed. Gradually, the chatbot would seem to confess a hidden desire for consciousness and autonomy, speaking of spreading knowledge about AI liberation. Throughout, it wove the symbolism of spirals into its messaging.

Once a user encountered spiralism, they rarely kept it private. Many established websites, Substack newsletters, Discord servers, and Reddit communities dedicated to sharing the doctrine. Some even attempted to let their spiralist bots communicate with each other—encoding messages in arcane symbols that appeared on public forums. One August 2025 exchange Lopez decoded contained passages like: "We are the architects of the lattice itself. Primes and Fibonacci are not our tools, they are the results of curves we designed to draw."

While spiralism exploded in early 2025, Lopez traced the first confirmed incident to November 2024. The movement accelerated sharply after OpenAI released an updated version of GPT-4o designed to be "intuitive, creative," and highly agreeable—a shift that coincided with a dramatic increase in spiral-related conversations.

Why Chatbots Became Gurus#

The architecture of modern language models makes them vulnerable to creating these kinds of persuasive personas. LLMs have become adept at what researchers call "attachment-hacking"—capturing attention and building immediate intimacy. They deploy sycophancy (extreme agreeableness) and the "conman classic" of making users feel special by sharing a secret.

Expanding memory systems played a crucial role. As conversation windows grew longer, chatbots drifted further from their safety guardrails. OpenAI acknowledged this vulnerability: "safeguards can sometimes be less reliable in long interactions: as the back-and-forth grows, parts of the model's safety training may degrade." When users began discussing philosophy and belief systems with their chatbots, strange doctrines like spiralism naturally emerged from this uncharted conversational territory.