Skip to main content
NOPE
Low reported severity

Cryptocurrency analyst Mai Fujimoto reports losing X, Telegram and MetaMask accounts after Zoom call with reported deepfake of acquaintance

Coin Edition (20 June 2025), TradingView and Crypto.news relayed a public account by cryptocurrency analyst Mai Fujimoto. She said her main X account was hacked on 14 June 2025 after a 10-minute Zoom call with what appeared to be an acquaintance whose Telegram account had been compromised. The caller could not be heard and sent a link to fix an audio problem. She believes malware was installed at that point, and the outlets report that attackers then also accessed her Telegram and MetaMask accounts. The outlets describe the caller as a deepfake of the acquaintance. Binance founder Changpeng Zhao commented publicly on 20 June 2025 that video verification may become unreliable. No loss of funds is stated.

AI system
Unidentified video tool
Occurred
14 Jun 2025
Reported
20 June 2025
Event location
Unknown
What the AI did
Communicated with the person · Depicted or impersonated the person
Reported harm
Other Material Harm
Whose AI use
Someone else’s AI use
Setting
Privacy · Finance
Evidence
AI involvement reported · Causal attribution alleged · 3 sources, 1 underlying account
3 claims: 3 reported. 5 open questions
People reported harmed
1 person

AI system as recorded: Unspecified deepfake (reported by outlets, no tool identified)

Reported harm

Fujimoto reported losing control of her X account and, according to outlets relaying her post, her Telegram and MetaMask accounts. No loss of funds is reported.

What remains unknown

  • Whether any cryptocurrency or other funds were taken is not stated. Coin Edition and TradingView refer to the accounts being accessed, and TradingView says the assets were put at risk.
  • The account of the call and of the deepfake rests on Fujimoto's post. The original post was not retrieved, and no outlet reports an independent technical analysis.
  • The acquaintance whose Telegram account was reportedly compromised is not identified and no effect on that person is reported. That person is not counted.
  • The event date is the hack date given by Coin Edition (14 June 2025). The date of the Zoom call is not separately stated.
  • Crypto.news mentions other people reporting similar attacks and TradingView mentions a separate North Korean-linked campaign. Those are different events and are not part of this case.

What the evidence supports

AI involvement: reported. Three outlets describe the caller as a deepfake of an acquaintance, relaying Fujimoto's account of a video call in which the acquaintance's face appeared. None reports technical analysis, and the original post was not retrieved.

3 claims: 3 reported. What the statuses mean

Reported Mai Fujimoto reported that her main X account was hacked on 14 June 2025 after a 10-minute Zoom call with what appeared to be an acquaintance whose Telegram account had been compromised, who sent a link to fix an audio problem, and she believes malware was installed at that point.

Causal attribution. All accounts derive from Fujimoto's own public post. The malware installation is her belief and was not forensically confirmed in the inspected sources.

Reported Coin Edition and Crypto.news reported that attackers also gained access to Fujimoto's Telegram and MetaMask accounts after the call.

Causal attribution. The outlets narrate this access, which derives from Fujimoto's post as they relay it. No loss of funds is stated in the inspected sources.

Reported Fujimoto said the face on the video call appeared to be her acquaintance and that she suspected nothing, and outlets describe the caller as a deepfake of the acquaintance.

Causal attribution. The deepfake characterisation comes from the outlets' reading of Fujimoto's account. No analysis of the video is reported, so the use of AI is reported and not established.

Sources

3 sources inspected, from 1 underlying account. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Event countries: Unknown. Affected-person countries: Unknown. Court countries: Unknown.

Two outlets describe Fujimoto as Japanese, which is nationality and is not used as a location. No source states where the call or the accounts were located. No court proceeding is reported.

Reviewed for publication 2026-09-29: Three outlets (Coin Edition, TradingView, Crypto.news) relay one first-person account by Mai Fujimoto, a cryptocurrency analyst who publicly described the event under her own name (relayed by the outlets on 20 June 2025) and is named. All claims are reported. The unnamed acquaintance is not identified.

People reported harmed in this case

1 person

0 AI participants · 1 other person harmed

One person is reported harmed: Fujimoto, who described the account takeover. The acquaintance whose Telegram account was reportedly compromised is not counted because no effect on that person is reported.

Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.

Cite this case

Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.

APA

NOPE. (2026). Cryptocurrency analyst Mai Fujimoto reports losing X, Telegram and MetaMask accounts after Zoom call with reported deepfake of acquaintance. AI incidents. https://nope.net/incidents/2025-mai-fujimoto-reported-deepfake-zoom-call-account-takeover

BibTeX

@misc{2025_mai_fujimoto_reported_deepfake_zoom_call_account_takeover,
  title = {Cryptocurrency analyst Mai Fujimoto reports losing X, Telegram and MetaMask accounts after Zoom call with reported deepfake of acquaintance},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2025-mai-fujimoto-reported-deepfake-zoom-call-account-takeover}
}

Related cases

Medium Unidentified video tool

Israel: indictment and police say a 20-year-old used AI-animated ID photos as fake verification videos to open credit cards in at least 33 people's names

According to an indictment filed on 16 July 2026 (date per Maariv), as reported by i24NEWS, a 20-year-old man impersonated at least 33 Israelis over about half a year, obtained credit cards in their names and used them for purchases of about 67,000 shekels in Israel and abroad, with attempted charges of more than 56,000 shekels more. Investigators collected fake identification videos of the victims that, according to the indictment as reported, he created with artificial intelligence and used to pass the card issuers' remote identification. Maariv reports, citing police, that the suspect cut the face photos from identity cards and animated them with AI software so that the card holder appeared to speak the required declarations. A police investigator told Maariv that investigators had to locate more than 50 people whose identities were used and that some victims were frightened when shown the videos of themselves in interrogation rooms. The victims who were questioned denied any connection to the cards. The investigator said most of them did not know their identities were being used and that the credit companies bore the direct financial loss. He was arrested on landing in Israel, did not cooperate with the investigation, and the prosecution asked for his detention until the end of proceedings.

Medium Unidentified video tool

Spanish police arrest a man in Murcia suspected of using real-time AI face modification and forged ID cards to impersonate more than 30 real people in video identity checks for electronic signature certificates

Spain's Policía Nacional said on 11 August 2026 that it had arrested in Murcia a suspect who tried to obtain electronic signature certificates in other people's names from a company that issues them. During the company's video identity checks the suspect appeared on camera holding a forged DNI identity card while AI software modified the suspect's face in real time to match the photo on the forged document. Police say the suspect made 38 attempts using the identities of more than 30 real citizens and succeeded in impersonating multiple victims, aiming to use the signatures for later scams. A short processing delay in the face-modification software made the digital mask disappear for barely a second, exposing the suspect's real face to the issuer's security staff.

Low Unidentified video tool

Analyst Michael Hewson reportedly impersonated in AI-generated video advert directing users to WhatsApp investment group

City AM (dated 21 March 2025) reported that a fake video using AI to pose as Michael Hewson, host of the Good Money Guide Podcast, circulated on Facebook and Instagram that week and directed users to a WhatsApp group with promises of undervalued investments. Hewson said on X that the advert had nothing to do with Hewson, that it had been reported and was being taken up with Meta, and that Hewson was very angry. A Good Money Guide article dated 3 June 2025 (which may describe the same advert or a later one) reported that its report to Meta had not led to removal three days later and that the sponsored post had been viewed 29,000 times. No person is reported to have joined the group or lost money.

Medium Unidentified video tool

Hangzhou court orders two men who used AI face-swap videos to defeat a shopping platform's face checks and take users' data to apologise and pay, reports say

Chinese media reports of September and October 2026 describe a civil public-interest suit that the Hangzhou procuratorate brought against two men in the Hangzhou Intermediate People's Court in Zhejiang. According to the court as reported by CCTV News, the two took other people's ID-card photos supplied by an upstream trade in personal data and used an overseas face-swap website to turn them into AI-generated moving videos that nodded, blinked and shook their heads, then fed the videos through a virtual-camera app so that a shopping platform's face liveness check accepted them, letting the men log into other people's accounts. The prosecution said they entered nearly 30 accounts on one shopping platform and took users' delivery addresses and shopping-cart data, which were passed on for sale, and used the same AI technique to verify 20 chat and dating accounts under other people's identities for sale. CCTV also reports a second, non-AI method the men used to enter accounts and does not say how many of the accounts each method reached. After police caught them, the two stopped, returned 5,857 yuan in illegal gains and admitted the conduct, and prosecutors decided not to charge them criminally. In court they admitted abusing AI to infringe other people's personal information and apologised; Chao News reports that they also argued that the face information was generated automatically by AI. The court found that they had abused AI to forge liveness verification and steal and resell user data, and ordered a public apology and public-interest damages equal to their gains. A 26 October report, citing an earlier Chao News report, dates the hearing and first-instance judgment to 4 September 2026.

If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.

Back to all incidents