A public archive of the Whistleblower Movement

关于 ghot.ai · 「爆料革命」知识档案 · 视频 / 帖子 / 榜单 / AI 问答

About · ABOUT BETA · Public preview Author's note · not an official position

I started following the "Whistleblower Movement" (爆料革命) several years ago. It produces a large amount of public content every day on X, Gettr, YouTube, and Telegram. I found myself scrolling timelines for two or three hours a day and growing more confused, not less—who said what, when they said it, whether outside sources corroborate it. ghot grew out of that. It is a tool for anyone following the "Whistleblower Movement" through its public record.

What you can do here what you can do here
01

A day or a week in 5 minutes

A daily AI summary distills the day into a 1–3 minute read, plus a weekly digest; go deeper by clicking any line back to the original posts and videos.

02

Read across languages

Key entries carry zh / en / ja machine translation, the original always beside the translation; follow external sources like the Trump and Takaichi teams too.

03

Search the record, jump to the moment

Video archives carry full transcripts — search a keyword to locate it, click a line to jump straight to that timestamp, no need to watch hours from the top.

04

One timeline across platforms

One person's accounts across X, Gettr and TruthSocial are gathered and merged into a single timeline — no more hopping between platforms.

A tool, not a publisher.

ghot is not a publisher. It does no original reporting and draws no conclusions. It does one narrow thing: gather, normalize, and index the public material—videos, posts, accounts, historical events—of the ongoing "Whistleblower Movement" so it can be searched, browsed, and questioned through AI. Every conclusion and every AI-generated summary traces back to a specific original source.

It is built for a specific reader: people who follow this movement day to day and want to make sense of it themselves—long-time readers, researchers, journalists, newcomers catching up. ghot compresses what would otherwise take hours—scrolling multi-platform timelines, finding the original video, lining up versions of the same statement—down to minutes.

AI inference ≠ self-declaration.

AI is used throughout the platform: labeling stance on posts and accounts, inferring the topic of a video, generating summaries. ghot keeps the two visually separate—any claim like "this account appears to support X" carries a confidence figure and sits next to that account's own self-description. Every inference knows that it is an inference.

We also do not put words in anyone's mouth. An anonymous account stays anonymous; an identity that has not been made public is not "inferred" into existence.

Multilingual.

Primary content is in Chinese. Important entries carry English and Japanese machine translations. Originals always sit beside translations; translations never replace originals.

— An independent researcher · New York · 2024 to present
Four principles product principles
01

Traceability over polish.

Every AI utterance links back to an original fragment. Summaries without a source are not shown.

02

Inference apart from self-declaration.

AI-inferred stance labels and account self-descriptions sit side by side. Confidence is visible.

03

Conflicts are not adjudicated.

When sources disagree they are shown side by side—no ranking, no "correct" label.

04

Originals are not lost.

Every ingest carries a hash and timestamp. Outbound links always point to the original platform.

Indexed videos
Hours of footage
Transcribed