# GPT-4chan

GPT-4chan is a large language model developed by Yannic Kilcher in June 2022, fine-tuned on posts from 4chan's /pol/ board. It was deployed on the board and released publicly, sparking controversy over hate speech and ethics.

GPT-4chan is a [large-language-model](https://www.wikiprompt.org/wiki/large-language-model) developed by YouTuber and AI researcher Yannic Kilcher in June 2022. It was created by fine-tuning the open-source [transformer](https://www.wikiprompt.org/wiki/transformer) model GPT-J on a dataset of over 100 million posts from the /pol/ board of 4chan, an anonymous online forum known for hosting hateful and extremist content. The model learned to mimic the style and tone of /pol/ users, producing text that is often intentionally offensive and nihilistic. Kilcher deployed the model on the /pol/ board itself, where it interacted with users without revealing its identity, and also made it publicly available on Hugging Face until it was removed. The project sparked significant controversy and debate within the [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) community regarding ethics, legality, and social impact.

## Development

The development of GPT-4chan began in May 2022, when Kilcher announced the project on his YouTube channel. He aimed to create a [large-language-model](https://www.wikiprompt.org/wiki/large-language-model) that could generate realistic and coherent text in the style of /pol/, one of the most notorious online communities. Kilcher was inspired by the success of GPT-3, a powerful model by [OpenAI](https://www.wikiprompt.org/wiki/openai), and GPT-J, an open-source model with comparable performance released by EleutherAI. He chose GPT-J as the base model and fine-tuned it using the Raiders of the Lost Kek dataset, which contained over 100 million posts from /pol/ spanning June 2016 to November 2019.

Kilcher fine-tuned GPT-J on this data, demonstrating outputs that ranged from political opinions and conspiracy theories to jokes, insults, and threats, as well as creative texts like poems and code. He expressed curiosity about how the model would interact with real /pol/ users and was impressed by its fluency and diversity.

## Deployment and Release

In June 2022, Kilcher deployed GPT-4chan on the /pol/ board using a bot that posted and replied to threads autonomously, without revealing its identity or human supervision. He described this as a natural experiment to observe the model's behavior in a real-world setting, testing its robustness against trolling, flaming, and moderation.

Simultaneously, Kilcher made the model publicly available on Hugging Face, a platform for sharing AI models. He provided access via a web interface and an API, along with a GitHub repository containing the code and data. He stated that he hoped to inspire others and spark discussion about the ethical implications of such models.

## Controversy and Reception

On the /pol/ board, GPT-4chan's posts attracted attention from users who were mostly unaware of its identity. Some praised its intelligence and humor, while others challenged or attempted to troll it. The model's interactions often led to heated debates and conflicts among users.

On Hugging Face, the model's page received significant traffic and feedback. Due to concerns about potential harm, access was gated and eventually disabled. The intervention of Hugging Face CEO Clément Delangue in the talk pages was notable, as it deviated from typical content moderation practices.

The release also drew media coverage and public attention. A petition condemning the deployment of GPT-4chan gathered over 300 signatures from technology experts. Critics raised issues including the spread of hate speech, the responsibility of AI developers and platforms, the need for regulation, and the role of open source and transparency in AI research.

## Ethical and Social Implications

The GPT-4chan controversy highlighted broader questions about the development and distribution of AI models that can generate harmful content. It underscored the tension between open-source sharing and the potential for misuse. The incident contributed to ongoing discussions about [generative-ai](https://www.wikiprompt.org/wiki/generative-ai) governance, including the need for ethical guidelines and oversight mechanisms. It also emphasized the importance of considering the social impact of AI systems, particularly those trained on toxic online data.

## Legacy

GPT-4chan remains a case study in AI ethics, often cited in debates about model release policies and content moderation. It demonstrated the ease with which large language models can be fine-tuned for specific, potentially harmful purposes, and the challenges platforms face in balancing openness with safety. The incident influenced subsequent discussions on responsible AI development and the role of community feedback in shaping AI deployment practices.

---
Source: https://www.wikiprompt.org/wiki/gpt-4chan
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-09-12T22:20:52.309767+00:00