New AI Disclosure Laws in 2026: A Simple Way to Sort Them

If you have tried to keep up with AI disclosure law this year, you have probably noticed the pattern: a new requirement lands every few weeks, from a different jurisdiction, aimed at a different kind of content, and every one of them gets covered as if it applies to everybody. Most of them do not apply to you. The trick is knowing which ones do.

Three disclosure laws landed within about five weeks of each other this September and August. None of them require the average person to label a work email. All three are worth understanding, because the pattern behind them tells you where this is heading.

Three Laws, Three Different Targets

California’s SB 1050, signed September 16, 2026, requires disclosure when advertisements feature AI-generated “synthetic performers,” a digitally created figure, voice, or representation that creates a realistic impression of a human performance without depicting any real, identifiable person. It applies to ads where that synthetic performer appears in a prominent role: demonstrating a product, delivering the primary commercial message, or serving as a testimonial. It takes effect January 1, 2027.

New York’s synthetic-performer law, enacted earlier in 2026, does similar work for advertisements in that state, requiring advertisers and agencies with actual knowledge of a synthetic performer’s use to conspicuously disclose it.

The EU AI Act’s Article 50, applying broadly from August 2, 2026, requires labeling of AI-generated or manipulated text when it is published to inform the public on matters of public interest, plus separate disclosure duties for deepfakes and chatbots. The full breakdown of who that duty actually reaches is in AI transparency and disclosure obligations in 2026.

Look at what these three laws have in common: every one of them targets content aimed outward, at the public, usually for commercial or informational purposes. None of them reaches into a private email, an internal memo, or a message to a named colleague.

The Three-Question Test

Instead of trying to track every new disclosure law as it lands, ask three questions about the specific piece of content in front of you.

Is it published to the public, rather than sent to a named person? An advertisement, a news-style article, a public statement: yes. An email to your manager, a client proposal sent to one recipient, a Slack message to your team: no.

Is it an advertisement? If AI helped write the copy or generate the imagery for something meant to sell a product or service, both the California and New York synthetic-performer laws, and increasingly other states, start to matter. Ordinary business correspondence is not advertising even when it is persuasive.

Does it impersonate or represent a specific human performance? Synthetic-performer laws are built around the specific harm of a fabricated person appearing to demonstrate, endorse, or testify. Rewriting your own words in your own voice, even with AI assistance, is not that.

If the honest answer to all three is no, you are very likely outside the disclosure duties that made headlines this year. If any answer is yes, that is exactly the moment to look up the specific law rather than guess.

Why This Keeps Multiplying By Jurisdiction

The pattern is not going to slow down. California, New York, and the EU each moved on disclosure in 2026 using different legal mechanisms, different definitions, and different effective dates, and other US states are working through similar proposals for employment and advertising contexts; see the state-by-state AI employment rules landscape for 2026 for how fragmented that picture already is even before disclosure rules are layered on top.

The honest takeaway is that “is AI disclosure required” is no longer a single yes-or-no question with one answer. It is a jurisdiction-by-jurisdiction, use-case-by-use-case question, and the only durable way to handle it is the three-question test above, applied fresh to whatever you are actually publishing, rather than a blanket policy borrowed from a headline.

For the fuller regulatory map beyond these three laws, AI transparency and disclosure obligations in 2026 is the place to start, and if your organization is still deciding on an internal disclosure norm separate from what the law requires, when you actually have to disclose AI use at work covers that distinct question.

Writing About This Without Overclaiming

Before:

Due to new AI regulations, all content must now include an AI disclosure statement.

After:

Starting January 1, 2027, ads using a synthetic AI performer in a lead role need a disclosure line. This does not affect our internal drafts or client emails; it only applies to advertising content.

The first version invents a blanket rule that does not exist and will cause someone to add disclosure language where none is required. The second states the actual scope, including what it does not cover, which is the detail people usually skip.

A Wrivio Context for summarizing a new disclosure law could say:

Rewrite this as a short, accurate summary of a new AI disclosure law for colleagues. Keep every jurisdiction name, bill number, date, and effective date exactly as written. Explicitly state what the law does not cover, not just what it does, and do not generalize a narrow rule into a broad one.

Press Ctrl+Shift+Space, paste your draft, and check the diff. Watch for the rewrite dropping the scope limits, since a summary that only states what a law requires, without stating what it excludes, is the most common way a narrow advertising rule gets misread as a blanket policy.

Common Questions

Do I have to disclose that I used AI to write an internal email in 2026?

No current law from California, New York, or the EU requires that. All three laws discussed here target public-facing content: advertisements with synthetic performers, or text published to inform the public on matters of public interest.

What counts as a “synthetic performer” under California’s SB 1050?

An AI-generated figure, voice, or representation that creates a realistic impression of a human performance in an advertisement, without depicting any identifiable real person, when it appears in a prominent commercial role.

When does SB 1050 take effect?

January 1, 2027, for advertisements using a synthetic performer in a prominent role.

Is the EU’s text-labeling rule the same as California’s advertising disclosure rule?

No. Article 50 targets AI-generated text published to inform the public on matters of public interest, while California and New York’s synthetic-performer laws target advertisements using AI-generated performers; they come from different legal systems and cover different content.

Where can I check the primary sources for these laws?

The Governor’s announcement of SB 1050 and the official Article 50 text are the primary sources referenced in this post.

Download Wrivio for Windows to draft the ordinary business writing none of these disclosure laws are actually aimed at, with a clear diff so you always know what an AI rewrite changed.