Commitments
ACATHRCA's Transparency Hub
What our products are, how they work, where they fall short, and how we keep them safe.
Chapter 01
Product report: ChatGiZa
What it is
ChatGiZa is a general-purpose AI assistant for writing, research, image and video generation, and coding. It is available on the web at chatgiza.com and as ChatGiZa Agent for Visual Studio Code.
How it works
ChatGiZa is built on large language models and generative media models, including models from third-party providers, combined with ACATHRCA's own instructions, tools and safeguards.
Known limitations
- It can state incorrect information confidently. Check important facts.
- Its knowledge of recent events may be incomplete.
- Generated code should be reviewed and tested before use.
Chapter 02
Data and privacy
How ChatGiZa collects, uses and retains your data is set out in the ChatGiZa privacy policy. We do not sell personal data.
Chapter 03
Safety evaluations
Releases go through the evaluations described in our Responsible Deployment Policy. We plan to publish summary results for major releases here.
Chapter 04
Policy enforcement
We enforce our usage policy through automated detection and human review. Aggregate enforcement figures will be published here once we have a full reporting period.