A new and dangerous AI threat for all Gmail users is seen in the wild
Google has implemented increasingly sophisticated protections against those who would compromise your Gmail account—but hackers using AI-driven attacks are also evolving. Here’s what you need to know.
The Latest AI-Driven Gmail Attack Is Scary Good
Sam Mitrovic, a Microsoft solutions consultant, has issued a warning after almost falling victim to what is described as a “super realistic AI scam call” capable of tricking even the most experienced of users.
It all started a week before Mitrovic realized the sophistication of the attack that was targeting him. “I received a notification to approve a Gmail account recovery attempt,” Mitrovic recounts in a blog post warning other Gmail users of the threat in question. The need to confirm an account recovery, or a password reset, is a notorious phishing attack methodology intended to drive the user to a fake login portal where they need to enter their credentials to report the request as not initiated by them.
Unsurprisingly, then, Mitrovic wasn’t falling for this and ignored the notification that appeared to originate from the U.S. and a missed phone call, pertaining to be from Google in Sydney, Australia, some 40 minutes later. So far, so relatively straightforward and easy to avoid. Then, almost exactly a week later, the fun started in earnest—another notification request for account recovery approval followed by a telephone call 40 minutes later. This time, Mitrovic didn’t miss the call and instead picked up: an American voice, claiming to be from Google support, confirmed that there was suspicious activity on the Gmail account.
“He asks if I’m traveling,” Mitrovic said, “when I said no, he asks if I logged in from Germany, to which I reply no.” All of this to engender trust in the caller and fear in the recipient. This is when things turned dark fast and really rather clever in the overall scheme of phishing things. The so-called Google support person informed Mitrovic that an attacker had accessed his Gmail account for the past 7 days, and had already downloaded account data. This rang alarm bells as Mitrovic recalled the recovery notification and missed call from a week earlier.
Googling the phone number he was being called from while speaking, Mitrovic discovered that it did, indeed, lead to Google business pages. This alone is a clever tactic likely to fool plenty of unsuspecting users caught up in the panic of the moment, as it wasn’t a Google support number but rather about getting calls from Google Assistant. “At the start of the call, you’ll hear the reason for the call and that the call is from Google. You can expect the call to come from an automated system or, in some cases, a manual operator,” the 100% genuine page helpfully informs the reader.
Lessons To Be Learned From This Gmail Hack Near Miss
Mitrovic did the right thing, or at least the next best thing to hanging up, and asked the supposed support guy to send an email confirmation—an email which arrived soon after, from a Google domain and looking for all intents and purposes genuine. AT this point he noticed the to field contained a cleverly disguised address that wasn’t actually a Google domain but could, once again, easily fool those not of a technical bent.
The real giveaway for Mitrovic, however, was when the caller said hello and after no response said hello again. “At this point I released it as an AI voice as the pronunciation and spacing were too perfect,” Mitrovic said.
It’s well worth reading the original blog from Mitrovic as it contains much more technical detail and detective work that I don’t have the space to cover in this report. Knowledge is everything, and the threat intelligence provided by this consultant is genuinely invaluable for anyone who might find themselves in a similar situation: forearmed is forewarned.
It’s almost a certainty that the attacker would have continued to a point where the so-called recovery process would be initiated, in truth this would be a cloned login portal capturing user credentials and likely the use of some kind of session cookie stealing malware to bypass two-factor authentication if that was in place.
AI deepfakes are not just used for porn and politics, they are used to perpetrate seemingly straightforward account takeovers such as in this case. Stay calm if you are approached by someone claiming to be from Google support, they won’t phone you so there’s a massive red flag right away, and no harm will come to you if you hang up. Use the tools at your disposal, ironically Google search itself and your Gmail account, to make checks during the call if you are concerned its could be genuine and ignoring it could cause harm. Search for the phone number, see where it’s really coming from. Check your Gmail activity to see what, if any, devices other than your own have been using the account. Take note of what Google says about staying safe from attackers using Gmail phishing scams. Most importantly, never let yourself be rushed into making a knee-jerk reaction, no matter how much urgency is injected into a conversation. It’s that sense of urgency that the attackers rely upon to swerve your normal good judgement and click a link or give up credentials.