AI-Generated Content

Some text in Spate is written by AI: Spate AI Assistant replies, report analysis and AI Insights, category descriptions, TikTok sentiment summaries, alert context briefs and parts of alert emails, and the headlines and "Spate POV" commentary in PowerPoint exports.

Metrics, charts, tables and figures are Spate data, not AI-generated.

How we label it

In the app and in emails: AI text carries a gray "AI-generated" label, or sits under a "Spate AI" or "AI Insights" heading. Sentiment tooltips end their source line with "Sentiment scored by AI". Boxes that send a question to Spate AI say so: the panel reads "The assistant is AI.", the homepage box invites you to "Ask Spate AI", and the Reports and widget boxes carry the "AI-generated" label.

PowerPoint exports: the cover and each slide with AI text carry a note. The file's properties (File > Info) list the keyword "AI-generated" and say which parts are AI.
CSV exports: the sentiment column is named "Sentiment Summary (AI-generated)".

Invisible watermark: AI text we generate now is produced with OpenAI's text watermarking turned on. It shapes word choices into a pattern only a detector can see, and adds no hidden characters. Text written before we turned it on, such as category descriptions, doesn't carry it, and passages under about 150 words, or heavily edited or translated text, may not carry a detectable watermark.

How to verify: Check the label, or the file properties of an exported deck. The watermark can only be checked with OpenAI's text detector, which isn't public. Qualifying organizations can apply at https://openai.com/form/content-provenance-api/. OpenAI explains how it works at https://help.openai.com/en/articles/8912793.

Questions: email support@spate.nyc. Regulators and researchers can use the same address.