Threat-Intelligence

The GTG-1002 Attack: The First Autonomous Cyberspace Campaign In November 2025, Anthropic revealed details of an unprecedented cyberespionage campaign, dubbed GTG-1002. This was the first time a case of an artificial intelligence (AI) agent orchestrating real-world intrusions with minimal human intervention was documented. A Chinese state-sponsored group manipulated an Anthropic Code Assistant to run approximately 80% of a multi-target hacking campaign autonomously. Instead of simply advising cybercriminals, AI took control of key phases of the operation, including: ...

Recent research has revealed the existence of four distinct threat activity groups that are leveraging a malware loader known as CastleLoader. This evidence reinforces previous assessment that the tool is offered to other cybercriminals under a malware-as-a-service (MaaS) model. The threat actor behind CastleLoader has been identified by Recorded Future’s Insikt Group as GrayBravo, previously tracked as TAG-150. GrayBravo’s Profile GrayBravo is a threat actor characterized by: Rapid development cycles. Technical sophistication. Responsiveness to public reports. An expansive and constantly evolving infrastructure. Tools and Frameworks GrayBravo’s toolset includes several key pieces of malware: ...

Two hacking groups linked to China have been detected weaponizing the newly disclosed vulnerability in React Server Components (RSC), known as React2Shell. The exploit was observed just hours after the existence of the flaw was made public, underscoring how quickly threat actors integrate new exploits into their campaigns. The React2Shell Vulnerability (CVE-2025-55182) The vulnerability in question is CVE-2025-55182, which has received a CVSS score of 10.0, indicating its maximum severity. This flaw allows unauthenticated remote code execution (RCE). ...

A joint investigation by Amnesty International, Haaretz, Inside Story and Inside IT has revealed that the human rights lawyer from Balochistan province, Pakistan, was the target of Intellexa’s Predator spyware. This incident marks the first time that a member of civil society in Pakistan has been targeted by this surveillance tool. The attack was carried out using a suspicious link sent by WhatsApp, which Amnesty International identified as an “attempted Predator attack” based on its technical behavior and characteristics. ...

Cloudflare has announced the detection and mitigation of a distributed denial of service (DDoS) attack that peaked at 29.7 terabits per second (Tbps), the largest ever recorded by the company. The attack, lasting 69 seconds, was launched by the botnet for hire known as AISURU. The AISURU Botnet: The Engine of the Attack Cloudflare identified that the attack came from the AISURU botnet, a cybercrime network that has been linked to numerous hypervolume DDoS attacks over the past year. The AISURU botnet is estimated to be powered by a massive network of between 1 and 4 million infected hosts worldwide. ...

The cybercriminal group known as Silver Fox has been identified orchestrating a “false flag” operation to imitate a Russian threat group. This tactic seeks to camouflage its attacks directed at organizations in China. The SEO poisoning campaign uses Microsoft Teams lures to trick unsuspecting users into downloading a malicious installation file. This file eventually deploys ValleyRAT (Winos 4.0), a malware associated with Chinese cybercrime groups. The activity has been taking place since November 2025. ...

Microsoft has quietly fixed a security vulnerability that has been exploited by multiple threat actors since 2017. The fix was included in the November 2025 Patch Tuesday updates. The vulnerability, tracked as CVE-2025-9491 (CVSS score: 7.8/7.0), is a “misinterpretation of the Windows shortcut file (LNK) user interface” flaw that could lead to remote code execution. Vulnerability Details (CVE-2025-9491) The vulnerability lies in how Windows handles .LNK files. The main problem is that a shortcut file can be manipulated to hide malicious commands from the user who inspects the file through the user interface. ...

Phishing kits typically have distinctive signatures in their delivery methods and infrastructure, making attribution easier. However, analysts have recently observed an overlap between two phishing kits such as Salty2FA and Tycoon2FA, marking a significant change that complicates detection. ANY.RUN observed a sudden drop in Salty2FA activity, followed by the appearance of Tycoon2FA indicators within Salty attack chains. Finally, unique payloads were detected that combined code from both frameworks. This convergence weakens kit-specific detection rules and gives threat actors more leeway to evade early detection. ...

Threat actors linked to the Iranian state have launched a new series of attacks against Israeli entities in various sectors, deploying a previously undocumented backdoor known as MuddyViper. The activity has been attributed to MuddyWater (also known as Mango Sandstorm or TA450), a hacking group allegedly affiliated with Iran’s Ministry of Intelligence and Security (MOIS). The attacks not only focused on Israel, but also on a technology company based in Egypt. Affected sectors in Israel include academia, engineering, local government, manufacturing, technology, transportation and public services. ...

North Korean threat actors responsible for the “Contagious Interview” campaign have flooded the npm registry with 197 additional malicious packages since last month. According to a Socket analysis, these packages have accumulated more than 31,000 downloads and are designed to distribute a variant of OtterCookie that combines features from BeaverTail and previous versions of OtterCookie. Infection Mechanism and Malware Capabilities The malware, once executed, performs various evasion actions, profiles the compromised machine and establishes a command and control (C2) channel. This channel provides attackers with remote shell and data theft capabilities, including: ...