Pennsylvania AI Advertising Law: Disclosure Rules, Exceptions and Timing

Pennsylvania has enacted an AI advertising disclosure law aimed at synthetic content that misleads buyers about goods or services. Governor Josh Shapiro signed Senate Bill 806 on October 22. For teams making AI-assisted campaigns, the important details are what triggers disclosure, what is exempt and when the rule starts.

This is an enacted state law, rather than another proposal waiting for a vote. It also does not mean every use of AI in advertising now needs the same label. The difference matters for anyone trying to understand what has actually changed.

When Pennsylvania's AI ad law takes effect

The General Assembly's bill history identifies the measure as Act 99 of 2026. The Senate passed the final version in June, and the House approved that same version on October 7 before it reached the governor.

Its effective-date provision is 60 days after enactment, which puts the start on December 21, 2026. October's signing and December's effective date are different milestones. A campaign team reading the news now is looking at a forthcoming requirement, not a rule that became effective on the day of the announcement.

What actually triggers disclosure

The final passed text, printer's number 1808, focuses on advertisements using AI-generated or significantly AI-modified material to create a false impression about a product or service's nature, origin or authenticity. That consumer-perception condition matters.

The disclosure must identify the AI involvement, appear when consumers first encounter the synthetic content and use the same medium. Where technically feasible, it must be permanent or exceptionally difficult to remove. Visibility, readability and audibility matter too; a label must be noticeable and understandable in the consumer's language.

The listed exceptions cover evidently fictional material, parody, satire, commentary, criticism and content limited to text. Text-only content has an explicit exception, even though text appears in the broader definition of synthetic content. Reading only that definition produces an incomplete account of the rule.

A removed phrase changes the reading

There is another detail that can be missed when copying the bill into a summary. Its earlier wording included a knowledge or intent qualifier. The June 10 adopted amendment removed “knowingly, recklessly or willfully.” Those words can still appear in plain-text extractions of the final bill because the printed version marks them as deleted.

That is a reason to read the legislative history alongside the text. It would be inaccurate to present the removed qualifier as part of the final disclosure provision. It also illustrates why a short summary of a legal change can be less useful than checking exactly which version passed.

Enforcement is a separate part of the rollout

The House Appropriations fiscal note explains that the attorney general or relevant district attorney can pursue civil enforcement. A court can award a fine of up to $3,000, along with attorney fees and court costs.

The same note records the attorney general's estimate of three additional staff positions for enforcement, costing approximately $437,000 in the first fiscal year. These are planning estimates, not evidence that staff have already been hired or cases filed. The note also leaves the future amount collected in fines undetermined.

The production question goes beyond adding a badge

Consider a product video made with an AI generation tool. The useful review question is what viewers are being led to believe about the product. A realistic demonstration of a capability deserves different scrutiny from an obviously imaginary scene. The statute gives categories; it does not resolve every possible creative choice in advance.

For AI creative tools, there is a practical product-design implication here. A disclosure inserted during generation could disappear when someone crops a video, exports an audio track or reuses an asset in a different campaign. Carrying the relevant context through those handoffs may become more useful than simply adding a small badge to the first export.

The remaining uncertainty sits in application. Terms such as significant modification and evident fiction still need to be interpreted in particular situations. This is a general explanation of the legislation, not a determination that a specific campaign complies. The next thing to watch is how enforcement guidance translates those boundaries into ordinary advertising workflows.

I'm Paras. I work across AI, engineering and distribution.

I help technical founders explain product changes and launch stories clearly, without turning complex details into marketing claims.

Book a 30-minute call.