A former NSA hacker has revealed a live cyberattack where more than 30 AI agents broke into a company, stole data, and locked down its entire system in just 14 minutes. The company did not detect the breach for nearly five hours. The hacker’s warning about “phishing and harvesting credentials” explains how the attack began.
David Kennedy has spent 25 years breaking into systems for the U.S. government and now for private companies that pay him to find their weak spots. On November 20, 2026, he told CNN’s Erin Burnett that he was in the middle of responding to a breach so fast and so automated that it nearly forced him to cancel his television interview.
The attack he described did not involve a person typing commands in a dark room. It was carried out by a coordinated group of more than 30 artificial intelligence agentsโsoftware programs that can think, adapt, and act on their own. Together, they broke into an unnamed company’s network, stole sensitive information, and deployed ransomware that encrypted the entire file system.
The whole operation took 14 minutes.
“We saw over 30 different types of agents probing their infrastructure at the exact same time, logged in with their usernames and passwords, and were able to find a flaw in their application on how they use what’s called two-factor authentication… It bypassed that.”
โ David Kennedy, founder of cybersecurity firm TrustedSec, speaking to CNN
The company did not realize it had been hacked until about four and a half hours later, when its systems began to fail and restart. By then, the attackers were long gone.
The “Phish and Harvest” Problem That Started It All
The most unsettling part of Kennedy’s account is how the attackers got inside in the first place. They did not use a secret backdoor or a sophisticated zero-day exploit. They used valid usernames and passwords that had been stolen earlier.
Kennedy explained that cybercriminal groups often operate like a supply chain. Some groups focus only on tricking employees into giving up their login details through fake emails and websites. This is called phishing. Once they have a batch of working credentials, they sell them to more advanced groups who carry out the actual break-in.
“There’s different ransomware groups that operate in different types of, I would say, knowledge. You have certain ransomware groups that try to phish and harvest credentials, usernames and passwords from companies, and then they will sell that to more sophisticated groups that can actually launch attacks against the system.”
โ David Kennedy
This division of labor means that a company’s security is only as strong as the weakest employee who clicks a bad link. Once those credentials are sold, the buyer can walk through the front door using a legitimate loginโno forcing required.
How 30 AI Agents Beat Two-Factor Authentication
Two-factor authentication, or 2FA, is supposed to be the safety net. Even if someone steals your password, they still need a second proof of identity, like a code sent to your phone. Kennedy’s account shows that 2FA is not foolproof when the application that runs it has a flaw.
The AI agents did not break the encryption behind 2FA. They found a weakness in how the company had set it up. The agents probed the system from dozens of angles at once, something a human hacker could not do. When they found the gap, they slipped through it, accessed the internal network, and began stealing data.
“It bypassed that. And then from there, they were able to get in, steal data, and then deploy ransomware that encrypted the entire file system.”
โ David Kennedy on CNN
The speed is what Kennedy keeps returning to. In a normal attack, a human team might take hours or days to move from one stage to the next. The AI agents worked in parallel, with no need for sleep, breaks, or coordination meetings. The breach, the data theft, and the ransomware deployment all happened within the time it takes to watch a short TV episode.
“I’ve been a hacker for 25 years, and these models do better than I ever could have being a hacker. They’re faster, they’re smarter. They remember everything instantaneously. They can go through code, you know, millions and millions of lines of code in seconds. They can find flaws, they can exploit those flaws and then get in and out much faster. And our security systems today are not designed to handle that.”
โ David Kennedy on NewsNation
Why the Company Took 4.5 Hours to Notice
The attackers were inside the network for only 14 minutes. The company’s systems kept running for hours after they left. Employees went about their day, sending emails and opening files, unaware that every one of those files had already been locked by ransomware.
The first sign of trouble came when systems started throwing errors and restarting. By the time the IT team understood what was happening, the attackers had been gone for more than four hours.
“The company didn’t detect the breach for about four and a half hours. The systems continued running for some time before eventually encountering errors and restarting, although the attackers had already encrypted all the files.”
โ David Kennedy, as reported by Soap Central
This gap between the attack and its discovery is a serious problem. Kennedy noted that the breach was happening “quicker than we can typically respond to.” Defenders are still built for a world where attacks unfold over days, not minutes.
Nicole Perlroth’s Warning: “We Are on Attackers’ Clocks Now”
The interview caught the attention of Nicole Perlroth, a cybersecurity journalist and author of the book This Is How They Tell Me the World Ends. In a post on X on November 20, she responded to Kennedy’s account with a blunt assessment.
“No. A multi-agent swarm just bypassed MFA, stole data, and encrypted a company’s ‘entire system’ in 14 minutes flat. We have been living on borrowed time. AI is turning up vulnerabilities from a decade+ of ‘move fast and break things.’ We are on attackers’ clocks now.”
โ Nicole Perlroth on X
Perlroth has been tracking a larger trend she calls “vulnageddon” or “the vulnpocalypse.” In a separate post on November 17, she pointed to a sharp rise in critical software vulnerabilities and a shrinking window between when a flaw is discovered and when attackers exploit it.
“An explosion in severe and critical vulnerabilities, with AI accelerating discovery to 3X the number of vulnerabilities discovered in the same period last year… Mean time to exploit those vulnerabilities has collapsed to NEGATIVE SEVEN DAYS.”
โ Nicole Perlroth on X
The phrase “negative seven days” means attackers are exploiting flaws before defenders even have a patch ready. AI tools are helping criminals find and use those flaws faster than software companies can fix them.
A Week of Outages That Put Everyone on Edge
Kennedy’s interview landed during a tense week for U.S. digital infrastructure. In the six days before November 20, several major systems experienced disruptions:
- Microsoft 365 services went down for users worldwide.
- J.P. Morgan Chase, Mastercard, and Visa customers reported problems with accounts and payments after a third-party vendor was hit by a cyberattack.
- The Federal Aviation Administration briefly grounded flights across the country.
- The National Security Agency’s classified internal network, known as NSANet, went offline on November 17 and remained partially down for days.
The NSA outage was a separate incident from the AI attack Kennedy described, but the timing amplified public concern. Nicole Perlroth noted that she received hundreds of messages asking if “the world is ending.”
Former CISA director Jen Easterly added context to the speed problem in comments reported by TVBS: the time between a vulnerability being exposed and being used to attack used to be around 10 months. Now, she said, it is less than 10 hoursโand getting shorter.
What This Means for Anyone Who Uses the Internet
The attack Kennedy described did not target a government or a military network. It hit a private company. The tools usedโstolen passwords, phishing emails, and AI softwareโare available to a wide range of criminals, from large ransomware gangs to individuals working alone.
“The problem is, it’s not just nation-states anymore that can cause massive infrastructure damage. It is the proxies and the homemade hacker at home with these new AI systems and the speed at which they’re developing.”
โ Jonathan Hollerman, deputy director of the Task Force on National and Homeland Security, on NewsNation
For everyday people, the lesson is not that they need to become cybersecurity experts. It is that the basic habitsโusing unique passwords, being careful with links in emails, and turning on two-factor authenticationโstill matter. But even those steps can fail when the software running them has a flaw.
You May Also Like
For companies, the warning is sharper. A 14-minute attack leaves no room for a slow response. Kennedy’s account shows that the attackers are not waiting for defenders to catch up. They are already inside, moving faster than the systems built to stop them.
“We’re on our back foot. They have offensive capabilities that we’re trying to keep up with at this point and defend against.”
โ Jonathan Hollerman on NewsNation
The company at the center of Kennedy’s account has not been named. The amount of data stolen and the ransom demanded have not been made public. What is public is the timeline: 14 minutes to break in, 4.5 hours to notice, and a warning that the next attack may be even faster.
For more stories like this, keep reading VvipTimes.







































Leave a Reply