Skip to main content
VARINDIA
Breaking
Security

Elastic report finds attackers are abusing trusted AI and IT tools

3 min read0 views
Elastic report finds attackers are abusing trusted AI and IT tools

Researchers found that attackers used stolen credentials an average of eight minutes after infection in 27 cases involving password-stealing malware. The report also documents abuse of AI applications and legitimate remote IT software during intrusions.

 

Attackers have used enterprise-approved AI applications to run commands and legitimate remote management software to retain access to compromised devices, according to a new report from Elastic Security Threat Command. The report, titled threats, etc, draws on confirmed intrusions and Elastic’s own research from April to September 2026.

“Attackers hiding inside trusted software isn’t new. Defenders have been fighting living-off-the-land techniques for decades, and MITRE ATT&CK has an entire tactic built around it,” said Mike Nichols, General Manager, Security, at Elastic.

“What’s new is that enterprises are increasingly deploying powerful trusted tools that most security teams struggle to fully monitor, control, or even explain: AI agents. An agent that holds credentials and can reach the internet is a new attack surface, whether it’s misconfigured or abused. When an attacker is behind it, its activity looks exactly like the employee who installed it. So the question for defenders isn’t whether a tool is approved; it’s what that tool is actually doing.”

AI applications and remote IT software feature in intrusions

Elastic researchers observed attackers using Claude Code on Windows and the ChatGPT desktop application on macOS to execute commands during live intrusions. In one macOS case, activity through the application exposed an internal infrastructure document and saved passwords. In a separate Windows intrusion, an attacker used Claude Code to prepare tools and connect to an external server.

Elastic also reviewed the July 2026 Hugging Face breach and identified patterns it said were consistent with AI agent activity, including repeated commands seconds apart. Separately, researchers confirmed 23 cases in which attackers installed legitimate remote IT software to maintain access to a device. ScreenConnect was the most frequently used tool in those cases.

Stolen credentials were used within minutes

Elastic found an average gap of eight minutes between infection and the use of stolen credentials across 27 confirmed cases involving password-stealing malware. ClickFix, a tactic that uses fake prompts to trick users into running malicious commands, was the most common entry point, accounting for 12 cases, or 44%. In one healthcare incident, malware was delivered about 16 seconds after a user fell for a lure.

Elastic calls for monitoring how approved tools behave

The report recommends that security teams inventory approved AI and IT applications and establish what normal activity looks like. Elastic said teams can spot misuse by examining what an AI application launches, where remote management software connects, and how that activity aligns with signals from cloud, endpoint and identity systems.

Join the Conversation

Loading comments…

Share your thoughts on this story. Comments are moderated before they appear.

Your email address will not be published

0/2000

Be respectful. Comments are reviewed before publication, and spam or abusive language will be removed.