
AI Training Data Transparency (CA AB 2013)
AI Training Transparency
A plain guide to the data behind our AI tools and what California AB 2013 asks us to share.
Last updated: April 25, 2026
Does Daylogue train AI models on user data?
No Daylogue model training. Daylogue does not use check-ins, voice text, journal entries, mood ratings, or reads to train its own models. Third-party handling follows the current service configuration, written terms, and subprocessor disclosures.
What models does Daylogue use, and where do they run?
- Anthropic Claude family: We reach these models through AWS Bedrock. Our setup does not let Anthropic view the input or output, and it does not keep model data.
- Deepgram: We use it only to turn voice into text. Our deal bars it from keeping or training on your audio and voice text.
Model work runs on our servers in our AWS account. One model does not share your data with another. We do not send it to a personal AI account.
Did Daylogue train any AI model itself?
Daylogue has not built or tuned a base AI model. We write our own prompts and use models from other firms. Since we do not train a model, we do not have a Daylogue training set to list under AB 2013. Anthropic and Deepgram share details about the data used to build their models:
- Anthropic training data summary: privacy.claude.com — model training data
- Deepgram model training disclosure: deepgram.com/privacy
How do users know they are interacting with an AI?
Daylogue tells you when a check-in uses AI. You see the note before the first text, voice, or SMS chat. We review that notice every three months. This also helps us meet EU AI Act Article 50 and California SB 1001.
What about telemetry, error logs, and analytics?
We remove personal content from system logs before we store them. Logs never train a model. Sentry error reports use a filter that strips private health data. Site data covers broad events and group totals only.
Changes to this statement
We review this page at least once a year. We also review it when we change an AI vendor, training rule, or model setup. We post major changes here and on the Trust & Security page.
Contact
Disclaimer
Daylogue is not therapy and is not a replacement for professional mental health care.