AI Disclosure

• Drafting and editing releases from your brief, URL or notes, at your direction.
• Scoring headlines and newsworthiness, AP-style checks and search/answer-engine (GEO) suggestions.
• A pre-screen that flags possible policy issues for our human editors.
• Translations you request, and personalized pitch drafts on Headline packages.
• A simulated AI visibility check: we ask an AI model how it would answer questions in your space and whether it would cite your release. It does not query live AI search engines.

• It never adds numbers, statistics, testimonials or quotes that aren't in your input. AI-suggested quotes are marked "Needs speaker approval" and can't be submitted until you attest the speaker approved them.
• It never makes the final editorial decision. A human editor reviews every release and every rejection.
• It is not trained on your content. Our AI provider processes it under no-training, zero-retention terms.

We record whether a release was AI-assisted. Because you approve every word, we don't force an "AI-written" label on your release. AI translations are labeled "Machine translated" unless a person reviewed them. AI-generated images of real people are not allowed, and other synthetic images must be labeled.

If text entered into an AI feature suggests someone may be in crisis, we show crisis resources (call or text 988 in the US, or 911 in an emergency) instead of continuing the task.