California’s new AI transparency law takes effect, requiring generative‑AI providers to embed provenance data and offer detection tools.

California’s AI Transparency Act officially went into effect today, mandating that generative‑AI services embed provenance information and provide users with detection tools.

Key Requirements of the Law

The legislation requires AI providers to attach a digital watermark or metadata that identifies the model, version, and the date the content was generated. Companies must also make an open‑source detection tool available so that platforms and the public can verify whether an image or text was AI‑generated.

Midjourney’s Current Status

Midjourney, a popular text‑to‑image service, does not yet embed a visible watermark in the images it creates. The company has said it is exploring technical solutions but has not announced a concrete rollout date.

Enforcement and Penalties

The California Department of Consumer Affairs will begin issuing fines of up to $2,500 per violation for non‑compliance. Repeat offenders could face higher penalties and be required to submit compliance plans.

  • Attach model and version metadata to every generated output
  • Provide a publicly accessible detection API
  • Display a clear notice to users before content is generated
  • Report compliance status annually to the state

The law also gives consumers the right to request removal of AI‑generated content that violates privacy or intellectual‑property rights, and it requires companies to maintain logs for at least 180 days.

Industry groups have warned that the rapid rollout could strain smaller startups, but regulators argue that transparency is essential to combat misinformation and deep‑fake abuse.

For a detailed overview, see TechTimes coverage of California AI Transparency Act.