AI disclosure law changes in 2026: a dated log of updates

This log records changes to the laws, regulations, guidance, standards and platform rules that govern labeling AI-generated content and telling people they are dealing with an AI, newest first. Each entry gives the date of the event, one or two sentences on what changed, and the primary source. For the rules themselves, start at the AI disclosure laws hub; for the rules that may apply to your content, use the AI disclosure laws finder. Accessibility law changes are in a separate log.

How this log is maintained

We follow our editorial policy: for each entry we read the primary source wherever one is published, such as the Official Journal, a legislature's own text or the regulator's page, and the link in each entry points to the best official source we could read. The date on an entry is the date of the event itself, such as publication, signature or the day a law takes effect, not the day we learned of it. Events we know only from press coverage are labeled as news reports. Forecasts are not entries: where a date is only expected, we say so and do not list it as fact. Where we derived a date from the text of a law rather than reading it, we say that. When we add or re-verify entries, the "Last reviewed" date at the top changes; if we correct an entry, we say what changed.

This log is for information, not legal advice. No tool can confirm that content or a site meets a law, and neither can this page.

Dates already set in published rules

These are dates written into texts that have been published. They are not predictions, but they can change, as the dates of the EU AI Act already changed once in July 2026.

One date is not in this table because we could not verify it: secondary sources report that South Korea's fines under the AI Basic Act are suspended for at least a year from January 22, 2026, which would put the earliest fine around January 2027. We did not read the decree that sets it.

2026

Entries marked "news report" record what a news source reported. We could not read the rule text itself.

October 13, 2026: Commission's page on the EU icons shows a new last update

The Commission's page on the EU icons for labeling AI-generated content shows a last update of October 13, 2026. As we read it on October 21, 2026, the icons are free to use without attribution, and the page says their use "does not establish legal compliance by itself". For the Partially AI-Modified icon it lists the use case "Authentic photographs of an empty apartment is furnished using AI". See our guides to accessible AI labels and virtual staging disclosure.

October 1, 2026: Connecticut's AI subscription notice and provenance sections take effect

Two parts of Connecticut Public Act 26-15 take effect. Section 1 requires a provider of AI technology under a subscription to give a Connecticut consumer a written notice of the key terms, and to receive the consumer's written acceptance, before it enters into or renews the subscription. Section 15 asks providers of generative AI systems with more than one million users a month to include provenance data in the audio, image and video content they create or materially alter. The Attorney General alone enforces both. See US AI disclosure laws.

September 30, 2026: California SB 1000 takes effect and rewrites the AI Transparency Act

SB 1000, an urgency statute, took effect and amended the California AI Transparency Act, which had become operative on August 2, 2026. After the amendment, the definition of a covered provider sets no user threshold, and a large online platform is one with 2,000,000 or more unique monthly users. The Act binds providers of generative AI systems and large platforms; in our reading it does not bind a business that only uses an AI tool to make its own content. The penalty is $5,000 per violation in a civil action by the Attorney General, a city attorney or a county counsel, and each day is a separate violation.

September 28, 2026: California approves AB 1609 on customer service chatbots

The Governor approved AB 1609. From January 1, 2027, a business with more than $500,000,000 in gross annual revenue may not represent that a customer service chatbot is human, and must disclose clearly and conspicuously that it is artificially generated and not human where a reasonable person is likely to believe it is a human. Only public prosecutors can enforce it, with penalties of up to $5,000 for a first violation and $10,000 for each later one, and the act creates no private right of action. See our guide to AI chatbot disclosure.

September 16, 2026: California approves SB 1050 on synthetic performers in ads

The Governor approved SB 1050, which adds Business and Professions Code 17610. An ad that prominently includes a synthetic performer needs a clear and conspicuous disclosure, in wording substantially similar to "this performance features a synthetic performer" or "no human performer is depicted". A violation is a violation of Section 17500 and may be enforced under Chapter 5 of the Business and Professions Code. The bill prints no effective date, so by the general rule it takes effect on January 1, 2027; that date is our derivation. See synthetic performer ads.

September 7, 2026: AccessibleEU says EN 301 549 v4.1.1 is not yet the legal reference

The European Commission's AccessibleEU center said version 3.2.1 of EN 301 549 remains the reference standard (AccessibleEU news). It matters here because Article 50(5) of the AI Act asks that the information conform to the applicable accessibility requirements; see how to make an AI label accessible.

August 18, 2026: IAB publishes version 2 of its AI transparency framework

The IAB published version 2 of its AI Transparency and Disclosure Framework. It is a voluntary industry framework for advertising, not a law.

August 2, 2026: EU AI Act Article 50 applies

Article 50 applies from this date to all in-scope systems, whenever they were placed on the market (Guidelines, paragraph 153). It covers telling people they are dealing with an AI, marking AI output in a machine-readable format, and labeling deep fakes and some AI-written text. Fines can reach EUR 15,000,000 or 3 percent of worldwide turnover; for small and medium-sized enterprises and small mid-caps the lower of the two applies. We found no enforcement action under Article 50 as of October 21, 2026. See Article 50 of the EU AI Act and our German guide to the KI-Kennzeichnungspflicht.

August 2, 2026: California AI Transparency Act becomes operative

Business and Professions Code 22757.6 makes the chapter operative on this date. It places latent disclosure and verification duties on providers of generative AI systems; the September 30 amendment above rewrote it. See AI image metadata for the markers involved.

July 31, 2026: Commission counts the Code of Practice signatories

The Commission's news item lists 95 signatories to Section 1 of the Code of Practice on transparency of AI-generated content and 192 to Section 2. The Code binds only those who sign it.

July 29, 2026: Germany's KI-MIG takes effect

The German KI-Marktüberwachungs-und-Innovationsförderungs-Gesetz (KI-MIG), issued on July 22, 2026 (BGBl. 2026 I Nr. 223), takes effect on this date according to the note on gesetze-im-internet.de, which also says the text there is not yet editorially processed. Section 2 names the Bundesnetzagentur as the market surveillance authority for the AI Act, except that authorities under state law supervise media service providers that use AI for journalistic or advertising purposes. Our German guide covers it: KI-Kennzeichnungspflicht.

July 24, 2026: Regulation (EU) 2026/1744 amends the AI Act

Regulation (EU) 2026/1744, dated July 8, 2026, was published in the Official Journal and entered into force on July 27, 2026. It adds Article 111(4): providers of generative systems placed on the market before August 2, 2026 have until December 2, 2026 to take the necessary steps for the marking duty in Article 50(2). Within Article 50 it replaces only paragraph 7. It also adds Article 99(6a), which gives small mid-caps the same lower-of fine cap as SMEs.

July 23, 2026: Amazon asks sellers to tag images of AI-generated people (news report)

CNBC reported on July 23, 2026 that Amazon told sellers on July 22 to add the XMP keyword contains-synthetic-performer to listing and A+ images of photorealistic people created entirely by AI. Amazon's own help page needs a Seller Central login, so we could not read the rule text, and whether older images must be tagged is unresolved. See platform AI disclosure rules.

July 20, 2026: Commission publishes its Guidelines on Article 50

The Commission published the Guidelines on the transparency obligations in Article 50 (C(2026) 5054 final). Paragraph 5 says they are non-binding. They explain who is a provider and a deployer, give examples of deep fakes, and say what a clear disclosure is. See who labels AI content.

July 15, 2026: FTC finalizes the TruHeight order on fake and incentivized reviews

The FTC finalized an order against Vanilla Chip LLC, which does business as TruHeight, and its principals, with a $4 million judgment partly suspended after $750,000 is paid. It bars misrepresenting that a reviewer exists and buying reviews for a sentiment. It is not an order about AI-written reviews; see AI-generated reviews.

July 9, 2026: Google Ads adds an AI content label setting

The Google Ads policy change log of July 9, 2026 records a setting that lets an advertiser label an asset as created or edited with AI. Google shows a disclosure in "How this ad was made" and overlays on ads in campaigns that target the EU, India and New York, and says the setting does not by itself settle whether an ad meets any law.

July 8, 2026: Commission finds the Code of Practice adequate

The Commission published its opinion on the Code of Practice on transparency of AI-generated content, finding it adequate for Article 50(2), (4) and (5). Following the Code is voluntary.

June 10, 2026: Final text of the Code of Practice

The final Code of Practice on Transparency of AI-generated Content was published. Its Section 2, Measure 1.1 asks signatories for accessible disclosure, such as high-contrast icons or labels and detectability by assistive technologies. The Code names no file format for marking.

June 9, 2026: New York's synthetic performer disclosure law takes effect

General Business Law 396-b, signed on December 11, 2025 as Chapter 617, took effect on the 180th day after it became law, which we calculate as June 9, 2026. It requires a conspicuous disclosure in an ad that includes a synthetic performer, where the person who creates the ad has actual knowledge, with civil penalties of $1,000 for a first violation and $5,000 for each later one. See synthetic performer ads.

May 27, 2026: Connecticut signs Public Act 26-15

The Governor signed the act, which includes the AI subscription notice, provenance data and employment decision notice sections described above. The sections took effect on October 1, 2026, and the employment duties apply from October 1, 2027.

May 19, 2026: FTC begins enforcing the TAKE IT DOWN Act

The FTC began enforcing the law that requires covered platforms to give people a way to request removal of intimate images shared without consent and to remove them, and known identical copies, within 48 hours of a valid request. It launched a complaint website and had sent letters to 15 named platforms the week before. It is not an AI labeling rule.

May 14, 2026: Colorado replaces its AI Act with SB 26-189

The Governor signed SB 26-189, which repeals and reenacts the Colorado AI Act of 2024 from January 1, 2027. We searched the signed act and found no general notice that a visitor is interacting with an AI; the notice that remains is tied to consequential decisions, and marketing and customer service triage are excluded.

April 28, 2026: China's regulator names three services for labeling failures

The Cyberspace Administration of China named CapCut (Jianying), Maoxiang and the Jimeng AI website for failing to label AI content and ordered interviews, corrections and warnings. The notice states no fine. It is the only labeling enforcement we found anywhere. See China and South Korea.

January 22, 2026: South Korea's AI Basic Act takes effect

The AI Basic Act is in force, with transparency duties in Article 31 for businesses that provide products or services using generative AI to users in Korea. Fines of up to 30 million won are possible under Article 43, and failing to indicate AI output is fineable only after a corrective order is ignored.

January 1, 2026: California AB 723 takes effect

Business and Professions Code 10140.8 requires a real estate broker or salesperson, or a person acting on their behalf, to disclose that a listing image was digitally altered and to link to the original. A willful violation is a misdemeanor under section 10185. See virtual staging disclosure.

January 1, 2026: Illinois requires notice of AI in employment decisions

775 ILCS 5/2-102(L), added by Public Act 103-0804, makes it a civil rights violation for an employer not to give an employee notice that it is using AI for recruitment, hiring, promotion and similar decisions.

January 1, 2026: Texas's AI governance act takes effect

HB 149 requires AI interaction disclosures only from governmental agencies and from providers of health care services or treatment (section 552.051). We found no chatbot or AI-content disclosure duty it places on private businesses in general.

Passed, but not law

Two New York bills have passed both houses. S.8451-B, the FAIR News Act, would require a label on content substantially created by generative AI, and A.3411-B would require a notice on a generative AI system's interface that its outputs may be inaccurate. On October 21, 2026 we found no delivery to the Governor and no signature on the Assembly's action history for either. They are not law and may change or be vetoed.

Earlier changes that still shape 2026

• December 22, 2025: FTC staff sent warning letters to 10 companies about the Consumer Review Rule, citing civil penalties of up to $53,088 per violation, and the FTC set aside its Rytr order (AI-generated reviews).
• December 11, 2025: New York signed General Business Law 396-b as Chapter 617.
• September 1, 2025: China's Measures for Labeling AI-Generated Synthetic Content took effect.
• June 12, 2025: Maine approved its chatbot disclosure law, Public Law 2025, chapter 294.
• May 7, 2025: Utah's SB 226 took effect, with a safe harbor in Utah Code 13-75-104 for AI that discloses at the outset and throughout.
• April 22, 2025: The FTC announced its final order against accessiBe, which bars claims that an automated product makes a website meet WCAG without evidence. It is why we never say that software alone makes anything meet a law.
• February 2024: Google's Merchant Center page on preserving metadata tags for AI-generated images is dated this month; see platform rules.

What to do with this log

Most entries move a date or add a duty for a narrow group, so the first step is to find out whether a rule reaches you. The AI disclosure laws finder does that from your audience and activities. If you publish AI-generated images, our AI image metadata checker reads the markers a file declares, and the AI label checker reads label wording on a live page.