Skip to content
useful information

Scams ChatGPT, Claude, and Gemini: when hackers steal passwords and MFA codes through fake advertisements

4 min read
découvrez comment reconnaître et éviter les arnaques courantes pour protéger vos informations personnelles et financières.

A wave of sophisticated scams is targeting online marketing professionals by impersonating the identities of ChatGPT, Claude, and Gemini. Hackers are deploying fake ads to steal passwords and two-factor authentication codes. This active threat hijacks advertising accounts, with malicious human oversight in real-time.

These scams are no longer based on classic malware but on very well-designed sites and fake login pages. Taking control of advertising accounts can lead to significant financial losses and compromise multiple clients simultaneously.

Here’s how these hackers exploit the popularity of the most well-known AIs to steal your access and how to effectively protect yourself.

The fake ChatGPT, Claude, and Gemini ads: a trap tailored for marketers

Cybercriminals stage portals that imitate the official interfaces of tools such as OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini. Each impersonation plays on enticing promises: expense audits, automated campaign planning, or simplified management of advertising accounts.

The victims receive an email inviting them to connect their account to supposedly benefit from ultra-personalized assistance. In reality, this click triggers a series of scenarios driven directly by hackers in real-time.

They can then control every step, forcing users to enter their passwords and various MFA codes multiple times while capturing this strategic data.

A stunning Google login simulation

Rather than redirecting to a real authentication page, these sites generate a fake browser window right within the page. This imitation reproduces the address bar, the padlock, and even the dark mode of the latest mobile browsers.

This trick works on all systems: Windows, macOS, iOS, Android. A real trust trap! The user, convinced they are on a legitimate site, unwittingly provides their credentials and codes.

Behind this facade, every password attempt is recorded, sometimes up to three attempts, allowing hackers to bypass apparent blocks or errors.

Human oversight: the new weapon for AI phishing campaigns

Researchers from Island.io have uncovered this infrastructure that combines open source technologies and human intervention. The platform centralizes attacks by operators who decide in real-time the next step in the authentication process.

After entering their credentials, a hacker chooses the requested MFA method: SMS, authentication app, push notification, QR code, etc. The victim remains on hold, often stuck on a loading screen while the intruder validates access.

This method makes phishing harder to detect. Automated systems block this type of attack little, as it resembles a normal human interaction.

A shared infrastructure for multiple scams

The same technical foundations are deployed for other scams, such as fake recruitment sites – some pages even mimic prestigious companies like Tesla or Ferrari. These traps exploit the trust and habits of professionals to use their credentials sometimes without caution.

A hacked advertising account can be used to finance fraudulent campaigns or resold. Recovery is often long and complex, as hackers can restrict the access of the legitimate owner.

The compromise of a single manager thus exposes multiple client accounts, their financial data, and ongoing campaigns.

How to detect and protect against fake ChatGPT, Claude, and Gemini ads

Extreme attention is now required to check the authenticity of requests related to artificial intelligence tools used in digital marketing. Here are simple yet essential actions to integrate into your habits:

  • Always systematically check official sites before clicking on a suspicious link received via email.
  • Check the real address bar of the browser and beware of nested windows or designs mimicking a browser.
  • Favor strong authentication via physical keys rather than passwords and one-time codes.
  • After a suspected attack, inspect all linked accounts for any unknown administrator or unjustified expenses.
  • Train your teams to recognize these threats and use free open-source tools to enhance security.

By 2026, this type of scam reflects a professionalizing phishing industry, relying on accessible tools and coordinated human labor. Staying informed and vigilant remains the best defense.

To understand in detail the stakes and solutions regarding these techniques, experts recommend regularly consulting reliable sources such as this article on fake ChatGPT ads or the detailed analyses of fraudulent sites.

How to recognize a fake advertisement imitating ChatGPT or Gemini?

Be cautious of email invitations requesting you connect an advertising account via a link. Always verify the real URL and be wary of pages that display an embedded authentication window.

What are the risks if my advertising account is hacked?

A takeover can lead to fraudulent campaigns charged to your account, loss of access, and compromise of multiple linked client accounts.

How to protect against these scams?

Use physical security keys, always verify links before connection, and never reuse the same passwords. Train your teams to detect these sophisticated attacks.

What to do in case of compromise?

Immediately report the compromise to your advertising platform, change all your access credentials, and check for the presence of unknown administrators. Monitor for unusual expenses.

Source: android-mt.ouest-france.fr

Bonjour, je m'appelle Gabriel, j'ai 27 ans et je suis expert en logiciels gratuits, libres et open source. Ma passion pour la technologie et le partage des connaissances me pousse à promouvoir des solutions accessibles à tous. Sur ce site, vous trouverez des ressources, des conseils et des tutoriels pour tirer le meilleur parti de ces outils puissants.