AI Output Disclosure Lint SB 942

It reads a JavaScript or TypeScript file and reports, with line numbers, where your generative-AI code drops a disclosure California SB 942 requires: metadata stripped on export, Content Credentials switched off, no AI label on the returned asset, a gated detection route, or licences handed out with no revocation path. Twelve rules, each tied to a duty in the statute.

Engineers who ship image, audio or video generation inside a US SaaS product that California residents can use. SB 942 reaches a covered provider, meaning a publicly accessible generative AI system with over one million monthly users or visitors. If you own the render pipeline, the upload path or the detection endpoint, these rules land on your files.

ESLint, TypeScript and the usual security scanners have no concept of a latent disclosure, so a call to withoutMetadata reads as ordinary image code to them. General chatbots do know SB 942 but most still cite the repealed January 1, 2026 start date rather than the August 2, 2026 date AB 853 set, and they cannot see your files.

Free and keyless: lint the file you have open, with all twelve rules, line numbers and the clause behind each finding. That finishes the job for one file. The paid tier is a different axis, scope and evidence: sweep every matching file in the workspace and export a dated Markdown and CSV report for counsel or a CI gate.

Outside privacy counsel reading one repository for AI-output disclosure gaps bills roughly four hundred to six hundred dollars an hour, and they still need an engineer beside them to find the render paths. The statute prices the other side too: five thousand dollars per violation, per day, each day counted separately.