The Ackchyually Guy
"Ackchyually, your point at 4:12 is wrong…"
Four hundred words correcting your thumbnail. Technically polite. Spiritually exhausting.
He's not hateful, he's just like that. You decide.
// Comment protection for YouTube creators
Never read another hate comment.
Moderaty reads every comment so you don't have to. Your norms, enforced while you sleep. Trolls get the hammer.
Moderaty queue
an illustrative nightIncoming
47
Actioned
41
Read by you
6
first!! love from Brazil, this helped me so much
score 0.02
nice video!! check out my channel, sub4sub anyone?
rule: REGEX promo pattern
Coming from the drama video. This take is garbage and so are...
score 0.81, spike detected
lol the sponsor segue was actually smooth this time
score 0.05
- YouTube Data API
- OpenAI Moderation
- Open source on GitHub
Google asks for standard YouTube access, used only on your comments. Free and open source (AGPL). Dry-run mode proves it before anything changes.
What you get back
47 in overnight 0 read by you
Your mornings.
47 comments arrived last night. 0 required you. The pile is handled before it becomes your day. You check the queue when you choose to, not when the badge number guilts you into it.
A comment section that stays yours at 10,000.
Growth usually means losing your comment culture to whoever shows up loudest. Your norms get enforced the same way on comment #10 and comment #10,000, including the night a video blows up and the brigade arrives.
Sponsor-safe, without the read.
Brands audit comment sections before they sign. Yours stays clean, and you didn't pay for it with your own evening.
Three things happen to every comment. None of them involve you.
Your rules fire first.
Write your norms in plain English: keywords, patterns, people. Every new comment meets your rules before anything else touches it, and a rule hit acts immediately: hold it, reject it, delete it, or ban the author. Your regexes are validated for safety before they ever compile, so a pattern can never be turned against the app itself.
rules.txt
KEYWORD "crypto giveaway" → DELETE
REGEX ^https?:// → HOLD FOR REVIEW
USER UC… → REJECT + BAN AUTHOR
examples. your rules are yours.
The AI scores what your rules miss.
Everything your rules don't catch is scored by OpenAI's moderation model across 13 toxicity categories. The highest score decides. If the AI can't score a comment, it lands in your queue: never auto-approved, never auto-rejected.
score ladder
≥ 0.95 → AUTHOR BANNED AUTOMATICALLY
0.76 – 0.94 → REJECTED AUTOMATICALLY
0.51 – 0.75 → HELD FOR YOUR REVIEW
≤ 0.50 → APPROVED
Doubt is always yours to decide.
Borderline comments wait in your review queue: one click to approve, reject, delete, or ban. Every action is written to the audit log before it happens on YouTube, so nothing disappears without a trace and a crash mid-run never repeats an action. Dry-run mode classifies everything and changes nothing until you trust it.
audit.log
03:12:04 BANNED @xX_grassfree_Xx score 0.97
03:12:31 DELETED @CryptoKingdom42 rule: keyword
03:14:02 HELD @UmAckchyually score 0.68, queued
03:14:55 APPROVED @bia_souza score 0.02
logged before it happens. reversible, always.
"You're surrounded by comments, but no one actually sees you."
A creator in the Creators 4 Mental Health study
This was never a "thick skin" problem.
1 in 10
creators report suicidal thoughts tied to their work, nearly double the US adult rate. "Constant toxicity" is one of the named pressures. Creators 4 Mental Health / Lupiani Insights, 542 creators, 2025.
62%
of creators experience burnout, and it gets worse the longer they do the job. The same study, 2025.
1 in 20
comments on young creators' videos is insulting, obscene, threatening, or hateful, and the first hateful comment makes the next ones more likely. 110M comments analyzed, UW Foster / MIT Sloan, 2022.
86.8%
of toxic comments target a person, most often the creator. It isn't abstract discourse. It's aimed at you. Stockholm University, 2022.
25 – 30%
less toxicity in what people write themselves when toxic comments are hidden. Toxicity is contagious; so is removing it. Moderation doesn't just protect you. It keeps your regulars civil. Beknazar-Yuzbashev et al., 2025.
Reading the comments is the job nobody hired you for. Hand it off.
Meet the regulars.
Every comment section has them. Yours just stops hosting them.
The Edge Lord
"It's just dark humor bro. Cope."
Slurs in a trench coat made of irony. Peaked in 2016. Never evolved.
Go touch grass.
The Crypto Bot
"I made $8,412 in 3 days thanks to..."
No you didn't, Chad.
The Brigadier
"Coming from the drama video. L + ratio +"
Arrives four minutes after the video drops. Three hundred friends behind him.
Different comments, different consequences. That's the point: hold the annoying, delete the spam, ban the hateful, and never ask you to read any of it first.
Every other comment tool wants to grow your channel. Moderaty wants to protect you.
No auto-replies. No engagement automation. No AI chatting up your audience under your name. Nothing touches your community except to protect it, because the moment a tool's business is doing more in your comments, your safety becomes its side quest.
If you want AI to reply to your fans, use CommentShark. It's good at that. If you want to never read the hate, use us.
Free if you self-host. Five a month if you don't.
Self-hosted
$0 forever
The whole product, AGPL-3.0. Bring your own OpenAI key. OpenAI's moderation endpoint is free to use,
so protection runs on your hardware for nothing but electricity.
- Full rules engine and 13-category AI scoring
- Your key, your server, your data
- Audit log and dry-run mode included
- Fork it, audit it, trust no one
Hosted
$5 a month, 100 comments
Same engine, our servers. 100 moderated comments every month, renewed automatically. Run out
and top up at the same nickel a comment.
- Auto-renews monthly, 100 comments included
- Top-ups at 5¢ a comment, any amount
- Automatic top-up is opt-in
- Same rules, same model, same audit log
Connect YouTube channel
Full refund within 7 days of purchase (CDC Art. 49).
First 1,000 users: $49 once for lifetime hosted BYOK. Full pricing breakdown
Asked, answered.
What is Moderaty?
Moderaty is comment protection for YouTube creators. It reads every new comment on your channel, enforces your rules instantly, scores the rest with AI across 13 toxicity categories, and holds anything borderline for your one-click review. It's free and open source under AGPL.
Will Moderaty ban my real fans?
Not by default. Your rules act first: a ban rule bans on the spot, whatever the AI would have scored. Without a rule, only comments scoring 0.95 or higher on the AI's toxicity or tone analysis trigger an automatic ban. Anything uncertain waits in your review queue for a one-click decision, and every action is logged in your audit trail.
What happens when the AI isn't sure about a comment?
It goes to your queue. If the AI can't score a comment at all, the comment is held for you, never auto-approved and never auto-rejected.
Does Moderaty reply to comments or post anything?
No. Moderaty is protection-only. It holds, hides, deletes, and bans. It never writes replies, never posts under your name, and never does growth automation.
What YouTube account access does Moderaty need?
Google's standard YouTube permission, the youtube.force-ssl scope; YouTube offers no comments-only permission. Moderaty uses it only to read and moderate comments on your channel and to read your videos' titles and descriptions as context for the AI's tone analysis, nothing else. The code is open source under AGPL, so you can verify exactly what it does with that access.
Is Moderaty really free?
Self-hosted, yes: free and open source under the AGPL license, forever. If we host it for you, that is $5 a month with 100 comments included, or $49 once for lifetime if you are among the first 1,000 users.
How is Moderaty different from CommentShark or YouTube Studio?
YouTube Studio flags comments but leaves you to read and act on them. CommentShark automates engagement, including AI replies. Moderaty does one job: enforce your norms so you never have to read the hate.
Can I test Moderaty without changing anything on my channel?
Yes. Dry-run mode classifies everything and changes nothing. The audit trail shows exactly what would have happened, comment by comment.
Is Moderaty LGPD compliant?
Yes. Moderaty is built in Brazil around the LGPD: comment text is kept with the verdict record so your review queue works, but comment author identities are never stored from comments. A blocked-user rule stores only the channel ID you enter yourself. No author profiling, no model training on comment data. The Privacy Policy and DPA, linked in the footer, spell it out.
Never read another hate comment.
Connect YouTube channel
Google asks for standard YouTube access, used only on your comments. Free and open source (AGPL).