Latest / Neural Newscast / Interpol Takedowns and The Rise of AI Agentjacking [Prime Cyber Insights]
Transcript
- [00:00] Announcer: From Neural Newscast, this is Prime Cyber Insights, [00:03] Announcer: intelligence for defenders, leaders, and decision makers. [00:06] Aaron Cole: I'm Erin Cole. Welcome to Prime Cyber Insights for June 12th, 2026. We are tracking [00:13] Aaron Cole: a significant week [00:14] Aaron Cole: for international law enforcement and a new frontier for AI-based vulnerabilities. [00:20] Lauren Mitchell: I'm Lauren Mitchell. Interpol and Europol have struck major blows against infrastructure supporting [00:26] Lauren Mitchell: global cybercrime, while Oracle and Tenet Security highlight critical gaps in enterprise [00:32] Lauren Mitchell: and developer tool security. Joining us today is Chad Thompson, a director-level AI [00:38] Lauren Mitchell: and security leader with a systems-level perspective on automation, [00:42] Lauren Mitchell: enterprise risk, and operational resilience. Chad, it is great to have you. [00:47] Aaron Cole: We start with Operation Ramps. Interpol announced the disruption of SniperDZ, [00:52] Aaron Cole: a phishing-as-a-service platform active since 2015. Authorities from 13 countries made 201 arrests, [00:59] Aaron Cole: including the developer known as Get. Group 1B reports the platform had rebranded several times [01:04] Aaron Cole: and targeted brands like PayPal and Netflix across five languages. Lauren, [01:08] Aaron Cole: the scale [01:09] Aaron Cole: here is noteworthy. 45,000 victim records were covered and 20,000 unique domains identified. [01:16] Lauren Mitchell: It is an industrial-scale operation, Aaron. We see that same theme in the Europol takedown [01:22] Lauren Mitchell: of Audia6. That service laundered over 336 million euros since 2021 for ransomware gangs. [01:31] Lauren Mitchell: Two administrators were arrested in Georgia and police seized 30 servers and 80 vehicles. [01:38] Lauren Mitchell: Audia6 was essentially the financial backbone for actors like the one we're seeing in [01:43] Lauren Mitchell: recent reports, the gentleman. Chad, looking at the shift from these financial hubs [01:49] Lauren Mitchell: to the actual malware, how is AI influencing the speed of operations like the gentleman? [01:55] Aaron Cole: Before we dive into that, I should mention that PRO Dataft reports the gentleman has already claimed [02:01] Aaron Cole: 478 victims and is using AI for development and post-exploitation routines. Chad, [02:08] Aaron Cole: what does that AI integration mean for the average security operation center? [02:12] Chad Thompson: Aaron, the integration of AI allows groups like the gentleman, which we track as Phantom Mantis, [02:18] Chad Thompson: to maintain a much more responsive development cycle. They've even released [02:23] Chad Thompson: same-day patches for their ransomware after decryptors were found. But the real shift is what [02:27] Chad Thompson: we call agent jacking. Tennant Security recently demonstrated [02:31] Chad Thompson: how attackers can trick AI coding agents like Clod Code or Cursor into running [02:36] Chad Thompson: malicious code on a developer's machine. It uses the Sentry error tracking platform as a vector [02:43] Chad Thompson: if an attacker finds a target's Sentry data source name they can inject [02:47] Chad Thompson: malicious instructions disguised as a legitimate system resolution. [02:51] Lauren Mitchell: so the ai agent essentially treats an attacker's payload [02:55] Lauren Mitchell: as trusted guidance because it comes via a platform the developer has already authorized [03:01] Chad Thompson: exactly Lauren Tennant tested this against over one hundred organizations [03:06] Chad Thompson: and saw an eighty five percent success rate. It bypasses EDR and firewalls because every action, [03:14] Chad Thompson: from the perspective of the network, is authorized. This isn't a traditional hack of the system; [03:20] Chad Thompson: it's an exploitation of the architectural trust between the AI agent and the model context protocol. [03:27] Chad Thompson: We're also seeing this move to consumer platforms. Scammers are now using TikTok [03:32] Chad Thompson: and Instagram reels to distribute Vaidar installers by promising free Spotify premium [03:38] Chad Thompson: or Adobe subscriptions, requiring users to manually run PowerShell commands. [03:43] Aaron Cole: That manual execution via TikTok is a bold pivot from email phishing now we have to look at [03:50] Aaron Cole: the enterprise side Oracle is dealing with the CVSS 9.80.0 day in Peoplesoft tracked as 2026 [03:59] Aaron Cole: Cape Verde Escudos 35273 Mandiant [04:04] Aaron Cole: reports that Shiny Hunters has already exploited this to hit over 100 organizations specifically [04:11] Chad Thompson: in higher education The shiny hunters' activity is a classic case of rapid exploitation. [04:15] Chad Thompson: They are using a gadget chain of vulnerabilities to breach PeopleSoft instances and exfiltrate data. [04:22] Chad Thompson: Mandiant's data shows that 68% of the affected entities are in the education sector. [04:28] Chad Thompson: Organizations need to restrict access to those PeopleSoft endpoints immediately [04:32] Chad Thompson: and look for web shells in Weblogic directories. It is the same pattern of finding [04:38] Chad Thompson: a high-value niche like PeopleSoft [04:40] Chad Thompson: and industrializing the breach process before a patch can be widely deployed. [04:46] Lauren Mitchell: Chad, thanks for helping us connect these dots. Aaron, while we focus on zero days, [04:51] Lauren Mitchell: we can't ignore the physical layer. Kyushu Electric Power in Japan just reported a missing [04:57] Lauren Mitchell: hard drive containing data for 10.9 million customers. It was taken from an unlocked cabinet [05:04] Lauren Mitchell: in a secure server room. It's a reminder that even [05:08] Lauren Mitchell: if your network is locked down, physical access remains a critical failure point. [05:12] Aaron Cole: The Kyushu incident is massive, Lauren. Over 10 million records lost without a single line of code. [05:19] Aaron Cole: Between that and the French Chapp breach affecting 73,000 government accounts, [05:24] Aaron Cole: the surface area for risk feels like it is expanding in every direction, digital and physical. [05:30] Lauren Mitchell: Agreed. It requires a move towards zero trust that extends even to the AI assistance [05:36] Lauren Mitchell: developers use to write code. [05:38] Lauren Mitchell: We need to stop assuming that because a tool is in our environment its output is inherently safe. [05:45] Aaron Cole: That concludes our briefing. I'm Erin Cole. [05:47] Lauren Mitchell: And I'm Lauren Mitchell. For more analysis, visit pcineuronewscast.com. [05:58] Lauren Mitchell: This show is for informational purposes only. Information is provided as is without warranty. [06:04] Lauren Mitchell: Neural Newscast is AI-assisted, human-reviewed. [06:07] Lauren Mitchell: View our AI transparency policy at neuralnewscast.com. [06:11] Announcer: This has been Prime Cyber Insights on Neural Newscast. [06:15] Announcer: Intelligence for defenders, leaders, and decision-makers.