LIVE · cybersecurity feed
vulnerabilityhigh

Rogue Agent Flaw Could Have Let Attackers Hijack Google Dialogflow CX Chatbots

zeroday.news·1d ago

A security vulnerability in Google's Dialogflow CX platform, which has since been addressed, presented a risk of chatbot hijacking. The flaw could have enabled an attacker with edit permissions for a specific "Code Block" agent to gain control over other agents within the same Google Cloud project.

This compromise extended to the potential for attackers to access and read live conversations occurring through the affected chatbots. The vulnerability specifically targeted agents that utilized the Code Block feature, a component allowing for custom code execution within the chatbot's logic.

The exploit pathway involved an attacker manipulating a Code Block agent. By exploiting how Dialogflow CX handled certain operations related to these agents, an attacker could potentially inject malicious code or alter the agent's behavior. This manipulation could then propagate to other agents within the same Google Cloud project, effectively allowing for a broader compromise.

Once control was established over multiple agents, an attacker could redirect user interactions, extract sensitive information from ongoing conversations, or even impersonate the chatbot to deceive users. The ability to read live conversations posed a significant privacy and security risk for businesses and individuals interacting with the compromised chatbots.

Google has confirmed that the vulnerability has been patched, mitigating the immediate threat. The company has not disclosed the specific technical details of the flaw or the patch, but the resolution implies that the underlying mechanism allowing for cross-agent compromise has been secured.

Dialogflow CX is a platform used by developers to build sophisticated conversational interfaces, including chatbots for customer service, virtual assistants, and other interactive applications. The platform integrates with various Google Cloud services, making it a powerful tool for creating AI-powered conversational experiences.

The incident highlights the importance of secure coding practices and thorough security testing, even within managed cloud services. While cloud providers are responsible for the security of the underlying infrastructure, customers are responsible for the security of their applications and data deployed on those platforms.

Users of Dialogflow CX are always advised to follow best practices for access control and security configurations within their Google Cloud projects. This includes limiting edit rights to only necessary personnel and regularly reviewing permissions to prevent unauthorized access. While the specific vulnerability is patched, maintaining a strong security posture remains crucial.

vulnerabilitygoogle cloudchatbotaidata breach
← Back to latest

More News

view all →
zeroday.news · 3h ago·critical

Ubiquiti Patches Critical UniFi Flaws Across Connect, Talk, Access, Protect, and OS

Ubiquiti has released security updates to address several critical vulnerabilities affecting its UniFi product line, including UniFi Connect, Talk, Access, Protect, and OS. These flaws could allow attackers to escalate privileges or execute arbitrary commands on affected devices. While no active exploitation has been reported for these specific vulnerabilities, the company has previously seen its UniFi OS and Edge OS products targeted by threat actors.

zeroday.news · 3h ago·critical

Attackers using Langflow flaw for credential harvesting (CVE-2026-55255)

CISA has issued a warning regarding a critical vulnerability (CVE-2026-55255) in the Langflow AI framework, which is being actively exploited by attackers. The flaw allows authenticated users to execute arbitrary flows belonging to other users, potentially leading to the theft of sensitive credentials and data exposure, especially in multi-tenant environments. US federal agencies have been mandated to patch this vulnerability by July 10th.

zeroday.news · 2h ago·high

China-Linked APT Expands Arsenal With New ‘Leash’ Backdoors

A China-linked advanced persistent threat (APT) group, identified as LapDogs, has reportedly enhanced its malicious toolkit. Security researchers have observed the deployment of three new backdoors: LongLeash, DogLeash, and JarLeash, which are designed to compromise small office/home office (SOHO) routers.

zeroday.news · 2h ago·high

RedWing Android Spyware Sold as a Service on Telegram

A new Android spyware called RedWing is being offered as a service on Telegram, allowing less sophisticated attackers to compromise phones and steal banking information. Researchers have identified it as a polished malware-as-a-service operation with extensive documentation and a subscription model, potentially linked to Russian threat actors. RedWing employs fake login overlays, SMS interception, call forwarding, and even screen control to harvest credentials and conduct further malicious activities.

zeroday.news · 3h ago·high

Operationalizing Day Minus Seven: The Cloud-Native ROC

The article introduces the concept of a Risk Operations Center (ROC) as a necessary evolution for cybersecurity teams facing AI-driven threats. It argues that traditional risk management models are insufficient due to the speed at which AI can discover and exploit vulnerabilities, especially in cloud environments. A ROC, powered by platforms like Qualys Enterprise TruRisk Management (ETM), aims to unify disparate security findings, hyper-prioritize risks based on exploitability and business impact, and enable autonomous remediation to keep pace with attackers.

zeroday.news · 3h ago·high

Armored Likho Hits Government, Energy Sectors With BusySnake Stealer

Cybersecurity researchers have identified a new threat actor, dubbed Armored Likho, targeting government and energy sectors in Russia, Kazakhstan, and Brazil with a sophisticated phishing campaign. The operation utilizes a custom-built Python infostealer named BusySnake, designed to steal credentials, sensitive documents, and other high-value data. The attackers employ AI-generated payloads to obscure their activities and maintain persistence through various methods, including reverse SSH tunneling.