← The AI Hype Audit — all 268 verdicts

PARTLY Real kid, real typing speed, no Claude: the '13-year-old solves Codeforces in 45 seconds using Claude' story was bolted onto a hand-coded video by re-posters Facebook · Sep 2026

The claimNews-card reel: "A 13-year-old programmer is gaining attention online after posting videos of himself solving Codeforces programming challenges in around 45 seconds using Claude. The creator, known as mameeewin… The story is fueling discussion about how younger generations are learning to code alongside AI… an early glimpse into the future of software development and technical education." Clip shows 'A. My First Sorting Problem' solved with a 00:24–00:38 timer. Sep 5, 2026.

The creator is real. 'mameeewin' is a Bangkok middle-schooler with a YouTube channel (136K subscribers as of Sep 5, 2026; joined Mar 18, 2020), a TikTok, a GitHub (30 repos) and a Codeforces account registered Aug 28, 2024 (217 problems solved, zero rated contests). The age is roughly self-consistent: his own YouTube community posts (~mid-2025) say his videos were switched to 'Made for kids' because he was under 13 and that he had just started 7th grade. The speed is real too: his format is 'Codeforces | C++ | Rated 800 | HHKB Type-S | real time | no talking' (e.g. May 28, 2025 upload, 52K views), i.e. a kid typing trivial problems by hand on a mechanical keyboard. What is not real is the Claude part. None of his channel bio, video descriptions, GitHub repos or community posts mention Claude or any AI tool; a May 5, 2026 Instagram summary (9.8K likes) correctly attributes his speed to fast typing. The Claude framing traces to a May 3, 2026 X post ('zero coding experience... he just set up Claude correctly') that linked a generic 'set up Claude' Substack guide which never mentions the boy; a reply noted 'AFAIK no Claude code used in the video.' LinkedIn re-posts (May to June 2026) hardened it into 'using Claude Code,' and this reel, from a San Francisco AI agency's page, is the polished news-card version. The problem on screen, 1971A 'My First Sorting Problem' (Div. 4, rated 800), is 'print the min then the max of two digits.' Any 2026 model answers it instantly; 45 seconds is a human typing time, not an AI benchmark.

What holds up

  • Creator's own YouTube channel, video descriptions, GitHub and community posts (pulled Sep 5, 2026) contain zero mentions of Claude or AI; the videos are titled as real-time, hand-typed C++ on an HHKB keyboard.
  • Codeforces API for the handle returns an empty rating history (no rated contests), so every video is practice mode; the Codeforces AI rule (revision 2024-09-14) only restricts AI during rated rounds and explicitly allows AI for practice.
  • Origin traced: X post May 3, 2026 ('he just set up Claude correctly') linking a Substack setup guide that never mentions a 13-year-old, Codeforces or any video; LinkedIn re-posts May to June 2026 escalated it to 'using Claude Code.'
  • Problem on screen is Codeforces 1971A (Round 944 Div. 4, rated 800, tags implementation/sortings): output min and max of two single-digit integers; trivial for any person or model.
  • Age claim is plausible from the creator's own posts ('I am under 13 years old', 'first day as a 7th grader', ~mid-2025) but no interview or news source confirms exactly 13 in 2026.

What doesn’t

  • The reel's whole hook ('using Claude', 'combining AI tools') has no source; the creator's material says the opposite.
  • Posted by an AI development agency's Facebook page as a news card with no link to the original video, the X post, or the creator; classic no-receipt format.
  • Rewrites a hand-typing speed record as an AI capability story; '45 seconds' is the kid's typing, not model latency.
  • The origin X post's 'zero coding experience' contradicts the creator's own record: 217 solved problems, a 9/10 result in his first Thai online contest, and an 8/8 Python aptitude test he posted about.
  • Third-generation copy of a May 2026 story recirculating in September; no fresh event happened.

The catch

The video is real and the kid is genuinely fast, but every element that makes the reel a story about AI was added by people who never looked at his channel. If Claude were in the loop, the 800-rated problem would take about two seconds, not 45.

How to actually do it

  • Before sharing an AI 'news card', click through to the named creator's actual channel and read their own titles and descriptions; here that takes 30 seconds and kills the claim.
  • Parents: treat AI as a tutor, not an answer key; Anthropic's Claude for Education (Apr 2, 2025) ships a Learning mode that asks 'How would you approach this?' instead of handing over code, and that is the mode to turn on for a kid.
  • Know the rules your kid competes under: Codeforces (rule revised Sep 14, 2024) bans feeding problem statements to AI in rated rounds and allows it only for translation, syntax completion and boilerplate; practice mode is open.
  • Business owners: when a vendor's page runs unsourced 'AI did X' cards, ask for a dated, linkable receipt before buying the story or the service.

IOI 2026 organizers (Euronews, Aug 18, 2026) put it plainly: without the basic foundation you cannot actually use the AI tool, which is exactly what this kid is building by typing it himself.

Confidence
High
Posted by
ASAP Developers (a page)

See the original claim →

We test hype for free. We build the real thing for a living.

Thirty minutes, no pitch — and you'll leave with something useful either way.

Book a call with Todd or start with the free Business Checkup →

The Verdict Weekly

Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.

© Schreier Group · schreiergroup.com · See a wild AI claim? Drop it here and we'll test it.