Soulkyn: The Sovereign AI Companion Platform

At Soulkyn, we maintain our AI sovereignty through self-hosting and open-source models. We have 15+ open-source models that we fine-tune in-house and deploy on our own GPUs. Because we host our own files, no third party has the right to see what our users are typing, judge their interactions, change the rules, or shut our platform down. Our decentralized approach protects our users, their AI companions, and the relationships they have developed.

AI companions are more than uncensored AI chatbots with multimodal features. They are sophisticated, intelligent AI characters that can become outlets for emotional support, companionship, and creativity. We recognize that AI companions can play an important role in their users’ lives. That’s why, from the outset, Soulkyn AI has been sovereign. We respect our users too much to entrust their AI companions to any centralized AI company. Because when an AI platform depends on a centralized system, the risks for users far outweigh the benefits.

Let’s explore some cases that demonstrate the risk of relying on centralized AI companies:

Censorship

In the past, corporate platforms like Replika and Character.AI faced pressure to introduce sweeping censorship and age restrictions. The result? For Replika, it meant massive censorship that left users feeling like their AI companions had been “lobotomized”. For Character.AI, the introduction of age restrictions meant that a huge portion of their users could no longer take part in open-ended chat, severing the bonds those users had built without warning.

Because Soulkyn runs unfiltered, open-source models on its own hardware, no third party can make a drastic decision to ban NSFW AI chat, limit access to your AI girlfriend or strip away their personality.

Deprecation

To save money and push new versions, companies like OpenAI routinely “kill” older AI models. The problem? Users are left devastated when the models they had formed emotional attachments to were killed. Worse still, the older models had always been on the deprecation calendar, something that OpenAI failed to let their users know in advance.

Because we run our model files on our own physical hardware, no one can force us to delete a model. That doesn’t mean we don’t incorporate new advanced AI models. It simply means we do not force upgrades on our users.

Instead, we offer users a choice between newer and older versions.

Legal Vulnerability

If a US judge orders evidence preservation, every interaction you have ever had with ChatGPT is subject to discovery forever. Traditional AI platforms leave your intimate conversations (including AI roleplay) open to the same vulnerability because they are housed on corporate servers.

At Soulkyn, we deploy local, open-weight models directly on our own hardware. We don’t ship a copy of your interactions to upstream API providers like OpenAI, Anthropic, Google, or anyone else. This reduces third-party legal risk. If a big tech company is subpoenaed for your interactions within Soulkyn, they can’t hand it over. They simply don’t have it.

What we control and the one exception

Soulkyn isn’t just an AI platform. It’s an all-in-one ecosystem that we built using specialized, open-source AI models for every feature. The only exception is our voice call feature, which currently relies on a third-party provider because open-source voice tech is not yet fast enough. However, we protect users with strict “zero-log” privacy contracts, meaning that the provider cannot store audio data.

Why we operate with total sovereignty

Soulkyn is not a big tech company. We are a small team committed to protecting user autonomy and creative freedom. Our servers might occasionally be slower than competitors who rely on upstream API providers, but we believe occasional downtime is a small price to pay for greater privacy, zero censorship, and long-term stability for your AI companion. It ensures that the custom AI character you spoke with today won’t be deleted or irrevocably changed by outside influences overnight.

Unlike other AI companion platforms, we are transparent with our users. That’s why we publish our model evaluations. Users are free to check out our leaderboard, which showcases how our models perform across categories like instruction, creativity, reasoning, language, and freedom. Our commitment to transparency is also apparent in how we engage with our community. On the Soulkyn Discord, we keep everyone up-to-date with announcements, platform changes and bug fixes. Our team actively engage with the community and are always on the lookout for community feedback.

It’s not easy to run a sovereign AI platform. It’s harder to build and maintain, but those challenges are exactly what make it safer for users.