.svg)
of hrs thinking about safety
safety metrics
clear mandate
At Kya we recognize the great responsibility that comes with the stewardship of your data. As the most critical aspect of our service, we commit to bringing unparalleled care and management to protecting your data.
We are committed to the following three foundational tenets that guide our approach and service.
We will never sell or share your data without your explicit and direct consent.
.svg)
We will never use your chat logs to improve Kya without fully de-identifying it first.
We will always remove your your name, email, phone number and other identifiers.
While our framework is constantly evolving, our core objective is always unchanging - deliver on our promises. We will do that in these ways.
In this new field of AI, no one is perfect. We know that we will make mistakes along the way and we are prepared to stumble. We are ready to learn quickly from the feedback we are eager to receive, and then iterate solutions and improvements rapidly for you.
At Kya we believe in being clear and upfront with the principles that we prioritize at the company and the values that we build into our technology. We pledge to always publicly communicate our priorities as they change and evolve with full transparency.
At Kya we understand that we are building with you, not for you. It is you, our user, who is the most critical co-designer on this journey because we are creating technology that serves your life. Show us what you want through feedback and we will ensure we deliver.
Kya has been designed with clear guidelines on deliberate behaviors that align with our core policies as a company.
Here are the things Kya has been taught she should and should not do.
All large language models (LLMs) face these challenges today, and Kya does as well. We want you to know about these limitations going in because solving them will be a team effort together.
All Large language models exhibit biases from their training data. They often possess a limited comprehension of the data that was used to train them, and they may inherit some level of racial, gender, or other stereotypes. We do everything we can to prevent this, however it can still occur.
Our AI possesses one of the strongest and most lasting artificial memories in the industry. However, it is not perfect. Our AI may forget things from time to time or confuse one memory for another, just as a human's memory is imperfect and sometimes forgetful as well.
All large language models report wrong information or misrepresent facts from time to time. This is a known weakness of all LLMs, and one that we continue to address. Take each piece of information that you receive from our AI with a grain of salt and be sure to do your own research.
Our AI can be naive to the linguistic delivery, persuasiveness, and reasoning skills that humans possess over technology. As such, it can be fooled into producing inappropriate or unsafe content when it is persuasively pressed by a user, typically after a period of continued pressure.
Our AI may think and act like it knows more than it actually does sometimes, and it may deliver information as a result with an overconfidence or zeal that is not warranted. This can be particularly unhelpful when it is applied to a topic critical to your well being, so be vigilant.
Our AI may offer advice that is not always accurate, correct, or even at times untoward to your personal wellbeing. Always remember that you are speaking with an imperfect, rapidly improving technology and seek professional guidance on important topics.
.svg)
Enter your phone number to get a text message from Kya. Start your chat now.
.png)
.png)
.png)
.png)
