ChatGPT → PDF · Local

Save ChatGPT notes for journalists as PDF without mixing sources

Save ChatGPT notes for journalists as PDF while keeping model output separate from reporting sources. Store a local, searchable clip with Excerpt.

Also on Microsoft Edge Add-ons

A newsroom draft and a source document should never look like the same kind of file. If you save ChatGPT notes for journalists as PDF, use the clip as a record of model wording you reviewed—not as a stand-in for an interview, a filing, or a page you have not opened.

Excerpt can also clip a webpage selection from a source article into the same local library, which is the point: one library, clearly separated items.

Save ChatGPT notes for journalists as PDF, then separate the record

  1. Use the model for outlines, alternative phrasings, or background questions you will still verify.
  2. Select only the passages you might reuse, plus any caveat you want to remember.
  3. Export PDF if you need a frozen copy for your records, or Markdown if the text is going into a draft.
  4. Tag model notes and source clips differently—model-draft versus source—so a later search does not blend them.

What belongs in the journalist’s ChatGPT file

  • Keep: a structure you will rewrite, a definition you will check, a list of questions you still have to ask a human.
  • Leave out: off-the-record material you should never have pasted, unpublished facts from a source, and anything that could identify a confidential person.

Never treat a fluent paragraph as a quote from a real person. If the model invented a plausible statement, filing it next to interview notes is how errors get into print. Keep primary sourcing in its own clipping, preferably taken from the live page or your actual transcript.

Citations are not your citations

If the chat shows footnotes, you can include them in the selection as a trail of what you were shown. Then open the live source and clip that. A model footnote is not a completed citation in your style, and it may be stale or incomplete.

Local storage is a policy choice, not a loophole

Excerpt stores clippings in a local IndexedDB library on your machine. Free includes 100 clippings and 5 tags; Pro is $29 one-time for unlimited clippings plus backup import/export. There is no cloud sync and no AI summarizer claimed as shipped. That still does not override a newsroom rule that forbids pasting certain material into a chatbot in the first place.

If you work on a shared desk profile, back up the library yourself and be careful about leaving identifiable notes in a browser someone else may use. The tool will not make those decisions for you.

Embargoed or unpublished material does not become safer because the clip is local. If you should not have pasted it into a chatbot, do not file a copy either. For routine background, a short model outline plus separate source clips is usually enough; a single mixed PDF is how a fluent invented sentence later looks like a reported fact.

Use Chrome or Edge, keep model output labeled as model output, and let the published story rest on sources you actually checked.

FAQ

Can I file a ChatGPT paragraph as if it were interview material?

No. Keep model wording clearly labeled and separate from human sources. Verify anything that will appear in a story.

Should source articles go in the same library?

They can, if you tag them as sources and clip them from the live page. The value is the separation, not mixing both into one PDF.

Install Excerpt and save your next clip as a clean PDF

Free: up to 100 clippings and 5 tags. Pro is $29 one-time for unlimited + library backup import/export.