Around 40% of the unique accounts amplifying the most-shared posts by news aggregator Mario Nawfal exhibit strong signals of being bots, fake profiles, or inauthentic accounts, according to a study by the Antisemitism Research Center of the Combat Antisemitism Movement (CAM).
The study, released on Tuesday, is titled “Bot-Like Amplification on X: Findings from CAM’s Study of the @MarioNawfal Amplifier Network,” and was conducted between August 29 and October 12, analyzing 10,015 unique X/Twitter accounts that amplified 230 of the most-shared original posts published by Nawfal.
Nawfal is a Lebanese-Australian entrepreneur, business influencer, podcast host, and political commentator who co-founded NFT Technologies and serves as CEO and founder of IBC Group.
Nawfal runs a news-aggregation account on the platform with roughly 3.95 million followers. Using public data from the official X API, 4,054 of these accounts showed strong bot-like signals, the report stated.
Nawfal accused of spreading hateful narratives about Ukraine, Israel
The analysis said that, “Nawfal’s team does not vet posts for accuracy, and since 2025 he has given long-form, largely unchallenged platforms to the Russian foreign minister and the president of Belarus, among other heads of state.”
The report added that “his content leans pro-Russian and hostile to Israel,” and noted that Conspiracy Watch describes him as “spreading false and hateful narratives about Ukraine, Jews and Israel.”
The report addressed the reach of the account, stating that “Elon Musk interacts with the account almost daily, regularly pushing its posts to millions of views.”
Combined with an amplifier network where roughly two in every five accounts behave like automated or fake profiles, this dynamic created “a ready-made distribution system that can push unverified claims and state-aligned narratives, including against Israel, to millions of users while appearing to reflect genuine public interest.”
The report noted that coinciding with the timing of the study, between October 8 and 11, Ukrainian drones struck three Yandex data centers in Russia, and Cloudflare radar showed automated traffic from Yandex’s network dropped by roughly 94%.
The report said that simultaneously, Israeli observers reported that accounts posting aggressive political comments in Russian and Hebrew fell silent, occurring ahead of Israel’s October 27 election and against a documented record of Russian bot operations targeting Jewish and Israeli audiences.
To evaluate the accounts, researchers used CAM’s detection model, known as X model v1.4.0, which measures each account against 32 signals across account behavior, profile characteristics, coordination patterns, and language.
The model is designed to ensure that “post content and ideology are not inputs, so an account is never scored as a bot for what it says.”
Bot-like activity differs from human-run account activity
Furthermore, the report noted that “any one measurable human behavior, such as original varied text, real conversations, or a daily sleep gap, pulls a borderline account out of the bot-like group.”
The signals of bot-like activity include specific automated patterns such as high posting rates and activity bursts, posting during odd hours, repetitive or duplicate text, default or placeholder avatars, digit-suffixed handles, abnormal follower patterns, and generic or repetitive replies.
Illustrating these behaviors, the report highlighted examples such as an account that accumulated 105,800 posts since October 2025 at a rate of about 280 posts a day using a placeholder avatar, another with 40,200 posts since April 2026 using a default avatar, and accounts created recently with minimal followers maintaining heavy, automated posting volumes.
Out of the 10,015 accounts evaluated in the final audit, 4,054 accounts (40.5%) displayed strong bot-like signals, 244 accounts (2.4%) showed moderate bot-like signals, 3,460 accounts (34.5%) exhibited some inauthentic signals without bot-like behavior, and 2,240 accounts (22.4%) were consistent with human activity.
The data also revealed that bot-like signals increase with user loyalty, rising to 54.6% among accounts that amplified three or more of the sampled posts.
The report also stated that “cross-account tests found a dense core of repeat amplifiers but no evidence of a centrally operated network: no copy-paste text across accounts, no sustained sub-minute synchronized reposting, and no clusters of accounts created in batches.”
Instead, the pattern indicated “a loyal, heavy-reposting audience in which a large share of accounts behave like automated ones.”
Pointing out that X’s authenticity policy prohibits “any activity that attempts to manipulate our platform or disrupt our services through inauthentic accounts, behaviors or content,” and that its platform manipulation policy forbids artificial amplification, CAM argued that enforcement is not keeping pace with platform rules.
CAM is calling on X to apply its platform manipulation policy to amplifier networks around high-reach accounts and to grant independent researchers necessary data access.