cybersecurity tips for Gmail users: Is your Gmail safe? Google exposes alarming AI hacking threats- here’s what you need to know
The company says urgent protections are required to stop bad instructions from entering E -mails and calendar invitations.
How does this Gmail AI attack work?
Security researchers showed how the computer pirates can use AI to enter Gmail. Hack begins with a fake calendar invitation that the victim does not have to accept. The user’s AI assistant checks the calendar and then monitors secret instructions to look at the e -mails according to a report made by Forbes without knowing the user.
Read also: iOS 26 has been started – see the new features that will come today
Eito Miyamura added an attack video to Gmail and sent it to X: “We bought a chatgpt to leak your special E -Post data.” “Everything you need? The victim’s e -mail address.” “AI agents like chatgpt, not common sense, as you follow your commands,” he warned, “We have managed to extinguish all your private information with your E -mail.”
Fast injection is a type of cyber attack where hidden commands are put into E -mails, files or invitations. Although AI systems cannot see users, they usually follow these instructions without questioning. This makes them very useful for pirates of computer, according to a report by Forbes.
Also read: shocking words about women and career goals days before Charlie Kirk’s death
What defenses did Google fulfill?
Google admitted that the threat was real and that it affects not only Gmail, but the whole industry. “It shows why it is important to develop solid guards against rapid injection attacks.”
“Our model training with problematic data significantly increased our defenses against indirectly fast injection attacks in Gemini 2.5 models,” Google said.
Also read: Is Tyler Robinson a Republican? Here is the political relationship of the suspect who was arrested by Charlie Kirk assassination
The company says Gemini 2.5 models already have better protection against indirectly fast injection, because they are trained with data from people trying to hack them.
Google also uses machine learning to prevent the bad demands in E -mails, attachments and calendar invitations. Classes may find harmful instructions and ignore them, which prevents AI from doing dangerous things.
Google argued that he has published registered machine learning models that can detect malicious demands and instructions in various formats such as E -mails and files ”. Gmail’s built -in guards, according to a report by Forbes, keep more than 99.9% of SPAM, identity hunting initiatives and malicious software.
How can people keep their accounts safe?
Although artificial intelligence defenses become better, user settings are still an important layer of protection. Google says you need to open the “known senders” setting on the calendar, so it doesn’t automatically invite people you don’t know.
“We found this as a particularly effective approach to help users to prevent malicious or spam events that appear on the calendar grill. Unless the user has made pre -interactions with the bad actor or changed the default settings, a certain calendar invitation would not be automatically descended.”
This step greatly reduces the probability of rapid injection attacks with bad events.
Experts also emphasize that you should be careful when using AI assistants. As the researcher Eto Miyamura shows, even strong tools such as Chatgpt can be used if the attackers seize an E -Posta address.
Gmail AI Hack reminds us how weak the new technology can be. As artificial intelligence becomes more common in daily vehicles, bad people benefit from a report by Forbes.
Google’s layered defenses and careful users will be important to stay in front of this growing cyber threat.
FAQ
What is Gmail AI Hack doing?
It directs AI assistants to leakage special e -mails by sending bad calendar invitations.
What to do first?
To stop receiving unwanted invitations, open the “known senders” setting in the Google calendar.
As a reliable and reliable news source
