Large language models (LLMs) and AI agents are becoming deeply integrated into web browsers, search engines and automated content-processing pipelines. While these integrations can expand functionality, they also introduce a new and largely underexplored attack surface.
One particularly concerning class of threats is indirect prompt injection (IDPI), in which adversaries embed hidden or manipulated instructions within website content that is later ingested by an LLM. This article shares in-the-wild observations from our telemetry, including our first observed case of AI-based ad review evasion.
Read more…
Source: Palo Alto Unit 42
Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox
Related:
- Beware of the license manager: how a Schneider Electric software vulnerability puts industrial facilities at risk
June 26, 2026
The CVE-2024-2658 vulnerability was discovered in 2024 within the FlexNet Publisher component of the Schneider Electric Floating License Manager. This software handles license management across various Schneider Electric products used for comprehensive industrial automation ranging from PLC programming to centralized control room implementation. This vulnerability is a CWE-427: Uncontrolled Search Path Element issue. It stems from a system ...
- Polymarket says hackers stole users’ funds
June 25, 2026
Prediction market giant Polymarket confirmed that hackers stole funds from an unspecified number of users after a third-party breach. In an X post on Thursday, Polymarket said that a compromise at a third-party vendor allowed hackers to inject malicious code into its website “for some users.” The company said it has “contained” the incident and is ...
- Self-destructing Mistic backdoor linked to access broker selling corporate footholds to ransomware gangs
June 25, 2026
A new self-destructing backdoor called Mistic used in intrusions since April appears to be linked to a criminal gang that compromises corporate networks and then sells that access to ransomware groups, according to security researchers. This backdoor, also tracked as MLTBackdoor, was first documented by Zscaler earlier this month, with the security shop suggesting the novel malware is ...
- Update Chrome to patch critical browser security flaws
June 25, 2026
Google released a security update for Chrome that fixes 18 vulnerabilities, including four rated Critical. There is no indication that any of these newly patched bugs are being actively exploited in the wild. The stable channel has been updated to 149.0.7827.196/197 for Windows and Mac and 149.0.7827.196 for Linux. The update will roll out over the coming days ...
- Almost half of ransomware victims have data stolen before they can even detect an intrusion
June 25, 2026
Criminals are getting better at hiding within their victims’ infrastructure, lurking and stealing files without triggering any alarms whatsoever. Earlier today, network detection and response experts ExtraHop released the “Global Threat Landscape Report”, based on a survey of more than 1,800 IT and security leaders worldwide. In it, it is said that roughly half (49%) of ...
- Global cyber strike disrupts SocGholish, Amadey, and StealC malware networks
June 24, 2026
Europol together with partners from across the globe today announces a landmark blow to cybercriminal networks as part of Operation Endgame, a sweeping international operation targeting the criminal infrastructure behind ransomware and malware like SocGholish, Amadey, and StealC. In coordinated actions over the past two weeks, key components of these malicious toolkits were dismantled as ...