Question
How is the AutoTableTop™ large language model (LLM) model trained? And what privacy controls does ThreatGEN have in place to ensure our data is not used elsewhere?
Versions:
- Single Use, Standard
- Plus
- Enterprise, Enterprise+, Enterprise+ on-premises
Answer
The AutoTableTop model is built on a foundation of OpenAI GPT models, which are then fine-tuned using a combination of public domain industry documents and ThreatGEN’s own test exercises.
No customer data or exercise transcripts are every received, saved, or viewed by ThreatGEN as all customer data and exercise data resides on the customer’s machine throughout their usage of AutoTableTop™. ThreatGEN does not have access to any user data, exercise data, settings, or prompts – prompts remain completely private and are never saved on ThreatGEN servers.
For more information on ThreatGEN and AutoTableTop™ data privacy, please see our policies page. We ensure data privacy through several measures:
- Exercise data resides only in the user’s local browser cache unless saved to a user-chosen location
- All data is encrypted in transit
- No data is ever saved on ThreatGEN’s servers
- User data is never used for AI training
Roadmap Q2 2025
AutoTableTop™ version 2.0, available sometime prior to the end of the second calendar quarter 2025, will have a document knowledge base list that can selected and integrated into an exercise (this would include many regional frameworks). This will be a set document list, but subscribers will have the opportunity to request specific frameworks and standards for inclusion in the knowledge base.
AutoTableTop™ Enterprise+ on-prem version 2.0 will have the ability to upload your own documents other than IR plans and network diagrams.
Copyright © 2025 by Derezzed inc. D/B/A ThreatGEN. All rights reserved.