Clean Paste AI

The Invisible Character Detector and Markdown Cleaner run in the browser: paste text, it flags and strips things like zero-width spaces (U+200B), the narrow no-break space ChatGPT's web UI inserts around numbers (U+202F), byte-order marks, and raw Markdown syntax (headings, bold asterisks, code fences), then hands back plain text. The vendor's privacy policy states this free path never sends pasted text to a server ("processed locally… never sent to our servers or saved anywhere"), which is consistent with it being a client-side JS text transform — a reasonable architecture for this task. There's no login or word limit on the free tools per the FAQ. This is the part of the product with a real, checkable technical basis.

Beyond the free cleaner, Pro credits unlock a 'Text Watermark Remover' and 'AI Humanizer' that send text to an unnamed third-party AI inference endpoint (the privacy policy doesn't name which one) to rewrite phrasing. Notably, the site is upfront about the limits of this: its FAQ and a dedicated blog post state plainly that Claude's official statistical watermark (per Anthropic's own August 2026 disclosure) contains no hidden characters and can't be removed by deleting characters — only a full semantic rewrite even attempts to touch it — and the terms of service explicitly disclaim any guarantee of bypassing Turnitin, GPTZero, or Pangram. That's a more honest framing than most "AI humanizer" products, which routinely oversell detector evasion.

Verified against the vendor's pricing page and its FAQPage/SoftwareApplication JSON-LD (https://cleanpasteai.com/pricing, checked 2026-08-30): free tier is $0 forever for the two local tools. Pro credits (10 credits = 1,000 words) come as one-time packs — 100/$5.99, 200/$9.99, 500/$19.99, credits don't expire — or monthly subscriptions: Starter 200 credits/mo at $5.99 ($0.30 per 1,000 words), Professional 420 credits/mo at $9.99 ($0.24/1,000), Advanced 1,000 credits/mo at $19.99 (~$0.20/1,000) — cost per word drops at higher tiers, which is a fairer curve than flat per-seat pricing. Cancellation is email-only ([email protected]), no self-serve button, confirmed in both Terms and the Refund policy.

Every page carries a banner reading 'Today Only: 50% OFF Annual Plans' with a countdown timer. Checking the numbers: annual price for every tier is exactly monthly price × 12 × 0.5 (Starter $5.99×12×0.5=$35.94, Professional $9.99×12×0.5=$59.94, Advanced $19.99×12×0.5=$119.94 — all match the pricing page precisely). That means the 'discount' is simply the standing annual-billing price, not a time-limited promotion, and the countdown showed the identical value (30:00:00) across multiple page loads taken several minutes apart during this review rather than counting down. It's a cosmetic urgency device on a price that doesn't actually change.

No legal entity name, business address, or team is published anywhere on the site — only a support email ([email protected]) and an About page describing the product in the abstract. Domain registration (RDAP, rdap.verisign.com, checked 2026-08-30) shows cleanpasteai.com was registered on 2026-08-20 via Namecheap — ten days before this review. Oddly, the site's own structured data claims the homepage was 'published' 2026-08-14, six days before the domain existed, which is either a copy-paste template default or a fabricated freshness signal; either way it shouldn't be taken as a real publish date. The three on-page testimonials (attributed only to "Publishing Editor," "Technical Writer," "Copywriter" with no names or companies) are unverifiable and not independently sourced. No coverage on Trustpilot, Product Hunt, or general search was found as of this review.

The sitemap (cleanpasteai.com/sitemap.xml) lists roughly 40 near-duplicate pages in English and Chinese — /clean-paste, /paste-clean, /clean-text, /text-clean, /unicode-remover, /unicode-cleaner, /chatgpt-text-cleaner, /gemini-text-cleaner, /humanize-ai-writing, /humanizer-ai-text, and similar keyword variants — each carrying the same tool widget, feature grid, FAQ, and testimonials with only the headline and meta description swapped per target keyword. This is a doorway-page pattern rather than distinct content, and it's the same content-farming approach that has drawn Google penalties on similarly structured sites.