Binance: CZ denounces “another attack”, this time on social networks

After the hack, the battle of messages. Binance would also have faced a coordinated amplification of hostile content on X, according to an analysis published by the account Ape (@immrape) and relayed on November 4 by Changpeng Zhao. Its author examines 11,700 posts from 1,083 accounts, published between October 30 and November 2.

The document describes repeated interactions and rumors accompanying the platform's announcements. These observations do not, however, make it possible to identify a sponsor or demonstrate an operation financed by a competitor, but the doubt remains.

Key Points

The author lists 11,700 messages and approximately 20.9 million impressions in his sample.

It classifies 3,354 posts as false, according to its own verification method.

Nearly 80% of the messages studied are responses to other publications.

The analysis identifies indicators of coordination, without establishing funding or common direction.

Binance: Analysis describes the amplification of rumors about

In its study entitled “Engineered Panic”, Ape explains that it used AI tools to identify accounts participating in discussions on Binance. It then selected those meeting certain criteria: repeated dissemination of refutable information, personal attacks, similar hostile messages or recurring interactions with other selected accounts.

According to his count, the publications studied generated 20.93 million impressions. This number measures views, not a number of distinct players. The author claims to have individually examined the 3,354 messages classified as false, divided into fifteen categories.

It also describes an evolution of the stories over the course of Binance's announcements: suspicions of internal operation, accusations of insolvency, then calls to withdraw funds as soon as withdrawals reopen. For him, this sequence suggests an organized amplification of distrust.

Changpeng Zhao relayed the work as “the other attack on Binance.” This reaction takes up the author's interpretation; it does not constitute an independent validation of its results.

Indications of coordination, but no established sponsor

To study the relationships between accounts, the author combines several signals: presence in the same discussion threads, mutual responses, subscriptions and close publications. Among the 641 most active accounts, it identifies 452 accounts with links using this method.

These comparisons nevertheless require caution. Users who share an opinion can comment on the same ads without receiving instructions. Likewise, increased activity after a hack does not, by itself, demonstrate artificial mobilization.

The sample by construction favors behaviors deemed problematic. It therefore does not represent all reactions to the incident. The author also acknowledges that he has no direct proof that a competing platform organized or financed the publications. His work thus raises a question about the circulation of rumors, without establishing the existence of a campaign driven by a company.