Now Anyone Can Build an Uncensored ChatGPT

FreedomGPT, an open source AI model unveiled last spring, is a ChatGPT-like assistant that its creators built to be uncensored. Now anyone can build on top of its platform with no filters: The API for FreedomGPT just dropped.

The idea behind FreedomGPT is that filters that prevent systems like ChatGPT from saying certain things are detrimental to freedom of speech and safety. They claim FreedomGPT “will answer any question without censorship, judgment or post-output bias.”

The FreedomGPT team decided to launch the API “in an effort (to) increase the utility” of its platform, according to its owner, Age of AI, an AI venture firm.

It is free to download and available for commercial use, though at the time of writing, the API is only accessible to users who have a Google account.

The API includes access to FreedomGPT’s Liberty, an uncensored large language model. Liberty has no post-processing features, with the team behind it claiming the model “will always produce answers based on its training data and not filtered after the fact.”

The API also includes an image generation model. Prior to the API release, users could access FreedomGPT via browser or download the desktop version for offline use on PC and Mac.

FreedomGPT provides access to third-party AI models for anyone to use regardless of technical ability. Some are censored, some are not, but both users and third-party systems are told to follow U.S. law “at all times” and not use hosted models for “unethical purposes.”

Related:ChatGPT Wannabes: Anti-woke FreedomGPT and Crypto’s Alex

AI safety through transparency

The launch of FreedomGPT comes at a time when the wider AI industry is trying to prioritize AI safety. Legislative approaches like the EU AI Act and in-house measures like OpenAI’s Preparedness team are some of the ways the industry is coming to grips with system security.

While lacking guardrails, the team working on FreedomGPT argues their method is about safety, too.

“Prioritizing unrestricted access to AI knowledge, (FreedomGPT) underscores the importance of privacy and unfiltered information,” the platform’s website reads. “Where traditional approaches risk introducing biases and skewed realities, FreedomGPT aims to democratize AI's power, paving the way for everyone to benefit from its advancements.”

The FreedomGPT view is echoed by tech billionaire Elon Musk, whose xAI startup launched an unfiltered chatbot called Grok to counter what he calls a “woke” ChatGPT. However, last week xAI was filed as a for-profit benefit corporation in Nevada, meaning it would have to produce a public benefit and operate responsibly and sustainably in addition to serving the financial interests of its shareholders.

Related:Meet Grok: Elon Musk's No Filter, Sarcastic ChatGPT Rival