Should captions censor profanity? Section 508 and YouTube's [ __ ]

Captions should show what the audio says. Section 508's guidance is that audible profanity goes into the captions as the exact words, a censored word is marked as censored, and no euphemism is swapped in. YouTube's automatic captions work differently by default: they replace potentially inappropriate words with [ __ ]. A burned-in caption from Sume shows whatever copy you give it, so the decision is yours and the check is easy to run.

Both rules below were read on 2026-10-03 from Section508.gov and YouTube Help.

What does Section 508 say about swearing?

Under Profanity, the page lists three rules:

• If the profanity is audible, include the exact words in the captions.
• If the audio is censored, indicate that, for example "I feel like (censored)."
• Never replace profanity with euphemisms or substitute words that were not used in the audio, such as writing "unalived" for "killed."

What does YouTube do by default?

YouTube's setting "Don't show potentially inappropriate words" replaces such words with an open bracket, two underscores and a closed bracket in automatic captions, and it is on by default. YouTube says it does not affect any audio tracks or manually edited captions, and it does not affect monetization status. The setting applies to automatic captions and auto-translated subtitles, and a channel can turn it off in YouTube Studio under Settings, Channel, Advanced settings.

So the same audio can read as exact words in a caption file you upload and as [ __ ] in the automatic track. That is a platform default, not an error in your file.

What ends up on screen with a burned-in caption?

The table compares the three situations. Platform rows come from the two pages above; the Sume rows come from the video captions docs.

How do I burn the wording I intend?

With script_text, Sume keeps speech-to-text word timings as the timing source and aligns the burned wording to your script, so the approved line is what appears. If you want a censored moment marked, use cues instead and write (censored) where the beep falls; cues skip speech-to-text and burn exactly that copy. script_text, words, cues and segments are mutually exclusive, so pick one per job.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: censored-cue-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/example/clean.mp4",
    "cues": [
      {"text": "I feel like (censored).", "start": 1.2, "end": 3.0}
    ]
  }'

Is this a legal or a brand decision?

Neither page is a statute for your video. Section 508 governs U.S. federal electronic content, and YouTube's setting is a channel preference. For your own ads and shorts it is a brand and platform-policy call. The useful habit is to decide it once, write it into the script, and burn the script, rather than let each platform's automatic track decide for you. Review whatever you burn before you render: a job that is accepted is billed at $0.20 for up to 60 seconds, so a typo in a cue is a re-run.