Benjamin L. Stewart

Properties1
tagsgenaidisclaimer

7 min read

GenAI Disclosure Statements

EU AI Act, Article 50 Disclosures

EU AI Act, Article 50 does not prescribe an exact sentence. It requires the disclosure to be:

  • Clear, distinguishable and accessible

  • Visible no later than the reader’s first exposure

  • Express about whether the text was AI-generated or AI-manipulated

  • Persistent when the content is shared or downloaded, where practicable

The Commission’s implementation guidance recommends plain language and distinguishes “Fully AI-generated” from “Partially AI-modified.” An optional EU AI icon may accompany the words, but an icon alone does not guarantee compliance. EU labelling guidance

For fully AI-generated public-interest text without human editorial review:

AI disclosure — Fully AI-generated: The text of this article was generated entirely using [AI tool/model] and has not undergone substantive human editorial review.

For materially modified or partially generated text:

AI disclosure — Partially AI-modified: Portions of this text were generated or materially modified using artificial intelligence.

If a human editor reviewed it and accepts responsibility, a transparent—but normally not legally required under this particular Article 50 provision—form would be:

AI-use disclosure: Artificial intelligence was used to assist in drafting or modifying this text. The final publication was reviewed and approved by [person/organization], which assumes editorial responsibility for its content.

The legal obligation is narrower than “any published text that used AI.” It concerns text published to inform the public about matters of public interest. The disclosure obligation does not apply when both:

  1. The AI-generated content underwent human review or editorial control; and

  2. A person or organization assumes editorial responsibility.

These requirements apply from August 2, 2026. Article 50 official guidance, European Commission implementation guidelines

LinkedIn Disclosures

If you publish the GenAI output essentially unchanged—without substantive fact-checking or editorial control—the best fit is “Fully AI-generated.”

Place this immediately below the LinkedIn article headline or at the beginning of the article, so readers see it before the generated text:

AI disclosure — Fully AI-generated: The text of this article was generated entirely using [AI tool/model] and has not undergone substantive human editorial review.

This aligns with the EU’s “Fully AI-generated” category for text created entirely by AI without human editorial control beyond prompting. The optional EU AI icon can accompany the words, but the visible text is the important part. European Commission labelling guidance

If you personally perform a substantive review—checking facts and sources, correcting problems, approving the substance and accepting responsibility—use this instead:

AI disclosure: This article was generated entirely using [AI tool/model]. I reviewed and fact-checked the final text and assume editorial responsibility for its publication.

In that second situation, Article 50’s mandatory text-labelling requirement may not apply because of the human-review/editorial-responsibility exception, but the voluntary disclosure remains transparent.

If text was generated by a human but reviewed by GenAI:

AI-assisted editing disclosure: I wrote this article. Generative AI was used only to suggest improvements to grammar, clarity, and word choice. I reviewed and approved all changes and remain responsible for the final text.

A shorter version…

Written by me and edited with AI assistance for grammar, clarity, and word choice. All changes were reviewed and approved by me.

LinkedIn currently describes Content Credentials for images and videos, not a comparable native label for ordinary article text, so include the disclosure in the article itself. LinkedIn also makes the member responsible for reviewing and complying with its policies regardless of whether content is labelled as AI-generated. LinkedIn User Agreement, LinkedIn Content Credentials

  • LinkedIn Agreement as it relates to publishing and AI

    • You remain responsible for everything you publish, whether written personally, generated through LinkedIn’s AI tools, or produced using an external AI service.

    • LinkedIn warns that generative-AI output may be inaccurate, incomplete, misleading or unsuitable. It instructs members to review and edit AI output before sharing it.

    • Labelling something “AI-generated” does not reduce your responsibility. The content must still comply with LinkedIn’s Professional Community Policies and applicable laws.

    • You must have the right to publish everything included in the post and must not infringe copyright, privacy, confidentiality or other rights.

    • LinkedIn may remove content or restrict accounts when content violates its agreement, policies or the law.

    • You retain ownership of your original content as between you and LinkedIn, but grant LinkedIn a broad, non-exclusive, worldwide, transferable and sublicensable licence to host, use, copy, modify, distribute, display and process it. Visibility settings limit how LinkedIn may distribute it.

    • Deleting content generally terminates that licence, subject to exceptions—for example, copies already reshared by others, earlier sublicences and required retention.

    • LinkedIn provides AI output and other content “as is” and puts responsibility on users to verify its accuracy and suitability. LinkedIn User Agreement

    For fully AI-generated material, whether it qualifies as “your original content” for copyright purposes is a separate legal question; LinkedIn’s agreement does not establish copyright ownership in AI output.

    LinkedIn Content Credentials

    LinkedIn is gradually implementing the C2PA Content Credentials standard:

    • It currently applies to qualifying images and videos, not ordinary LinkedIn article text.

    • If uploaded media contains cryptographically signed C2PA metadata, LinkedIn displays a visible Content Credentials icon.

    • Opening the credential may reveal whether AI was reportedly used, the originating application or device, the creator, the credential issuer and its timestamp.

    • The credential records the media’s claimed provenance and modification history. It does not establish that the content is factually accurate.

    • LinkedIn cannot detect or label all AI-generated media. A missing icon therefore does not mean that AI was not used.

    • The feature is being rolled out gradually and may not yet be available to every member. LinkedIn Content Credentials

    Practical implication

    For an AI-generated LinkedIn article, place a visible textual disclosure inside the article itself. Do not rely on LinkedIn’s Content Credentials, because those currently concern embedded credentials in images and videos. Regardless of the label, fact-check the article, confirm that you have rights to all included material, and treat yourself—not LinkedIn or the AI provider—as responsible for what is published.

More Disclosure Examples

  • Option 1 (AI used only for editing): “AI disclosure: This article reflects my own ideas, experiences, and conclusions. AI was used only to assist with editing, grammar, clarity, and readability. Any errors or opinions remain my own.”

  • Option 2: “AI disclosure: I wrote the original draft of this article. AI was used to help improve organization, wording, and readability. The ideas, examples, and interpretations are my own.” - Best for LinkedIn

  • Option 3: “AI disclosure: AI was used during the drafting process to help generate questions, challenge assumptions, and suggest alternative phrasings. All arguments and final editorial decisions are my own.”

  • Option 4: “AI disclosure: Portions of this article were generated with AI based on prompts and source material that I provided. I reviewed, edited, and approved all AI-generated text before publication. The overall argument and responsibility for the content remain mine.”

  • Option 5: “AI disclosure: Unless otherwise noted, this article was written by me.” “The section titled insert section title here was generated with AI from prompts and materials I provided, then reviewed and edited by me for accuracy, clarity, and consistency.”

  • Option 6: “AI disclosure: This article combines human writing and AI assistance. My contributions include the topic selection, analysis, examples, and conclusions. AI assisted with editing and, where noted, drafting specific passages. I reviewed every section before publication and take full responsibility for the final content.”

  • For a standing disclosure you use consistently: “AI transparency statement: I believe AI should augment, not replace, human thinking. In my writing, AI may be used for brainstorming, organization, editing, or drafting selected passages. The ideas, professional judgment, and final responsibility for all content are mine.”

  • And a concise voice-aligned version: “Transparency note: I use AI intentionally, as an editor, thought partner, and occasional drafting assistant, not as a substitute for original thinking. Every article reflects my own professional judgment, and I take full responsibility for its content.”

Graph View

Backlinks