As of March 2026, the US does not yet have a federal law that universally mandates visible watermarking on all AI‑generated synthetic content, but there are regulatory efforts and proposals moving in that direction.

At the federal level there has been legislation proposed that would require tools that generate synthetic content to provide the option for machine‑readable provenance metadata that identifies content as synthetic, and to make that metadata meaningfully tamper‑resistant, but it has not yet been enacted as a comprehensive mandate across all AI outputs. Some parts of that proposal would take effect a couple of years after enactment if passed.

There are also state‑level laws and regulatory trends relevant to watermarking and labeling of synthetic content. For example, New York State passed laws requiring clear disclosure when ads use AI‑generated synthetic performers, and that goes into effect in mid‑2026. That’s not a blanket watermark rule for all content, but it does require transparency when generative AI is used in advertising to avoid consumer deception.

Some U.S. states have considered or introduced bills (such as in California) that would push for watermarking or provenance standards for AI‑generated content, and major tech firms have expressed support for labeling standards, but no nationwide law currently obligates all AI systems to watermark every piece of synthetic content in a visible form.

At the same time, industry self‑regulation and platform practices (such as voluntary watermarking by some AI tools or metadata tagging standards like those used by C2PA) are becoming more common, reflecting a trend toward greater transparency even without a federal mandate.

In short: there isn’t yet a U.S. federal mandate forcing all AI‑generated content to carry a visible watermark, but regulatory trends, executive directives, and state laws are moving in that direction, particularly where transparency and consumer protection are at stake.