AI Usage Policy

Last updated June 22, 2026

1. Purpose

Column, operated by Pattern Build, LLC — the owner of the Column trademark and intellectual property — uses artificial intelligence to help agencies draft professional content in the style of the clients they represent. This policy explains how our AI works, what we commit to, and what we expect from you. It exists because responsible use of AI — especially when content is published in a real person's name — protects you, the people you write for, and us.

2. How Column's AI works

Column generates drafts ("Output") using third-party large language models accessed through their APIs. When you generate Output, your relevant inputs and the applicable Voice Profile are provided to the model as context, and the model returns a draft to your workspace. A Voice Profile is a set of instructions and stored context that guides style; it is used as prompt context at generation time. Column does not train or fine-tune models on your data (see Section 4).

3. AI limitations — read this

AI Output can be wrong. You should assume that Output may:

• Contain factual errors or "hallucinations," including invented statistics, quotes, events, names, or citations.
• Misstate facts about real people or companies, which—if published—could be false, misleading, or defamatory.
• Be non-original or resemble existing material, and may not be protectable by copyright.
• Be outdated, because models have knowledge limits.
• Reflect biases present in training data.

Output is a draft and a starting point, not a finished, verified, or publication-ready product, and not professional, legal, financial, or other advice.

4. Our commitments

• No training on your data. We do not use your Customer Content, Voice Profiles, transcripts, or Output to train, fine-tune, or improve any AI model — ours or any third party's — on any plan. The only exception is if you expressly opt in to a specific program; absent that opt-in, the answer is no.
• Our AI providers are bound too. We use model providers under contracts that prohibit training on your data and that apply zero- or short-retention terms, so your data is not used to improve their models either.
• Workspace isolation. We do not use one customer's data to generate Output for another. Your Voice Profiles and content stay in your workspace.
• No sale of your data, and no use of your content for advertising.
• Transparency and provenance. We are building toward machine-readable marking of AI-generated Output and related provenance signals to support emerging transparency requirements (including the EU AI Act).

5. Your responsibilities

By using Column's AI features, you agree to:

• Keep a human in the loop. A qualified person must review, fact-check, and approve all Output before it is published. Do not publish unreviewed Output. This step is required by the Terms of Service.
• Verify facts. Independently confirm any factual claim, statistic, quote, citation, or statement about a real person or organization before publication.
• Have the rights and consents. Only generate content for Represented Persons you are authorized to write for, and only from materials you are permitted to use, consistent with Section 4 of the Terms of Service.
• Disclose AI use where required. Where law or a platform requires disclosure that content is AI-assisted or AI-generated, you are responsible for making that disclosure. Maintain records of human review where you rely on an editorial-responsibility exception.
• Use lawful inputs. Do not upload unlawfully obtained recordings, biometric identifiers, or third-party confidential information.

6. Prohibited uses

You may not use Column's AI features to:

• Impersonate or create content in the name or voice of any person without authorization, including any person other than a Represented Person you are authorized to write for.
• Generate or publish content you know or should know is false about a real person or organization, or content intended to deceive about its authorship in an unlawful way.
• Create fake reviews, testimonials, or endorsements, or content that misrepresents a genuine experience or a material connection (this can violate FTC rules).
• Produce synthesized audio or video replicas of a person's voice or likeness. Column generates text in a writing style; it is not for voice cloning or deepfakes.
• Generate unlawful, harassing, hateful, or infringing content, content that violates third-party rights, or content that violates a platform's terms.
• Attempt to defeat safety features, extract another customer's data, or reverse-engineer the models.

Violations may result in suspension or termination under the Terms of Service.

7. Publishing and platform integrations

Column produces drafts for human review; it does not auto-publish on your behalf without your approval. Where Column integrates with third-party platforms (such as LinkedIn), it does so only through those platforms' official APIs. We do not perform or authorize scraping or unsanctioned automation, and you are responsible for complying with each platform's terms when you publish.

8. Changes

We may update this policy as the Service and the legal landscape evolve. We will post the updated version with a new date and, for material changes, provide additional notice.

9. Contact

Questions about this policy: support@hicolumn.com.