AI Disclosure and Communications

Two important pieces of legislation came into effect on August 2, 2026, that are part of a wave of rules being passed in the United States and around the world. The EU AI Act and the California AI Transparency Act have the goal of addressing the malicious use of generative AI for deepfakes and misinformation, and require transparency around the use of the technology.

The laws differentiate between AI companies generating content and organizations publishing content. Although the first phase of the California law is primarily focused on AI companies, in the broader legislative context, the responsibility for transparency will also fall on the shoulders of publishers – and that means communicators. The EU AI Act already requires that AI-generated content be disclosed.

In a new Tauth Labs white paper, The EU AI Act, Content Authentication, and AI Disclosures, we go into some detail about what the Act does and does not cover. It’s important to note that we are not lawyers and that the rules are subject to change. The starting point for companies and communicators is to conduct a legal or compliance review of what these laws will mean for your company or organization.

Here are some broad-brush takeaways:

1. AI companies that generate content need to disclose that it is AI-generated. 

2. If your company is using AI to generate or manipulate content and you publish it – whether it is text, an image, or video, you will need to disclose it.

3. If you are producing content that is targeted for European audiences or that will reach people in California, New York, or other states/countries with transparency legislation – you will need to comply with that legislation no matter where you are based.

4. Under the EU AI Act, if you are a news publisher using AI to generate content, but have a human in the loop with editorial responsibility, there is an exemption from disclosure.

5. If content is obviously AI-generated, disclosure may or may not required, depending upon the legislation.

6. Deepfake content must generally be disclosed as AI-generated; specific types of harmful content are banned outright.

7. For slides and charts generated in the normal course of business using AI, disclosure may not be required.

8. If you are using AI to generate images or content that looks authentic, the use of AI will need to be disclosed.

In a world in which AI touches almost everything, there will be lots of grey areas around these rules. For example, what level of AI manipulation of content triggers the new laws? The rules do have litmus tests, but they will not always be clear for every scenario.

One thing required today under the EU AI Act is that disclosure of AI content must be both human- and machine-readable. A label is not enough, and content authentication, a valuable tool to address machine-readable disclosure requirements, must also include a visible ”AI Generated” label. This label can be automatically generated when authentication is built into workflows.

In practice, AI-generated articles, releases, images, or content will require disclosure. If you are using AI to create realistic images and are using them for social media, for content that targets European audiences this needs to be disclosed. Companies and communicators cannot rely on disclosures that may be built into the tools you use; as a “publisher” of that content, you have a disclosure obligation. 

With human-produced content a feature of the communications industry (with AI assist), a takeaway for communicators is that you need to review from a disclosure perspective all types of content being produced for business or clients, including images on social media. With disclosure requirements in effect today – this must be a priority.

Content authentication provides a way to address AI disclosure requirements for people and machines, and incorporate “AI Generated” labels. It is a Swiss Army Knife technology (See our 10 Use Cases). It builds trust into content, makes it more discoverable, and in this context, it is a compliance tool. 

In the new regulatory landscape, communicators need to be thinking about AI disclosure. How it works in practice is likely to evolve over time. There's likely a long and winding road to travel on this journey, but our call is that in a sea of slop, the trust signal embedded into content through authentication, which will include AI transparency disclosures, will differentiate high-quality content generated by communicators, companies and AI.