Enterprise Technology Signals

SGLang has two unauthenticated RCE issues and one unauthenticated path traversal tied to specific configs and endpoints.

Netskope adds AI risk attributes to Cloud Confidence Index and describes prompt/response guardrails for SaaS apps with AI agents.

JupiterOne launched AI Attack Surface Management (AI ASM) and Unified Vulnerability Management (UVM) to provide relationship-aware asset and vulnerability context.

Arctic Wolf launched Aurora Mobile Threat Defense for iOS and Android, adding real-time mobile phishing and unsafe network detection plus Threat Intelligence Plus and Concierge enhancements.

Vendor blog highlights a governance gap for AI used via MCP, APIs, and CLI, citing a report that 92% of orgs lack MCP policies.

A local privilege escalation flaw in Linux kernel versions 4.17+ can let an unprivileged user gain root access. The issue, CVE-2026-31431 (“Copy Fail”), involves algif_aead/AF_ALG page-cache writes and targets in-memory setuid binaries.

Infoblox completed its acquisition of Axur and will add DRPS URL scanning with AI, feeding DNS blocking and takedowns into Infoblox Threat Defense.

Synack made Sara AI Pentesting generally available, pairing agentic AI with human validation for continuous security testing.

Tencent’s Weixin 2025 Brand Protection Report cites AI detection, user reports, and brand partnership outcomes, including $430M+ recovered enforcement value.

Intercom [email protected] used a preinstall flow to download Bun, harvest GitHub credentials, and use them for npm spread.

Huawei launched Xinghe AI Network Security Agentic SOC with Sensing, Analysis and Enforcement agents for automated detection and response.

Fortinet will return to the World Economic Forum Annual Meeting on Cybersecurity 2026 in Geneva, citing participation in public-private cyber initiatives.

NSS Labs publishes two AI security white papers and a new AIPS test methodology focused on runtime guardrails like prompt injection and unauthorized output prevention.

Fortinet’s 2026 Global Cybersecurity Skills Gap Report links talent shortages, board AI risk gaps, and breach costs to ongoing cybersecurity breaches.

Unauthenticated access to DRC INSIGHT COS /v0/configuration lets same-network users modify config, enabling data exfiltration or disruption (CVE-2026-5756).

Censys added integrations with AI, SIEM, and SOAR platforms, including Cisco Splunk SOAR and ES, Microsoft Sentinel, and Google SecOps.

Ollama’s model quantization engine has an unauthenticated remote information disclosure flaw in CVE-2026-5757 that can let an attacker with model upload access read and exfiltrate server heap memory.

Overview A remote code execution vulnerability has been discovered in the SGLang project, specifically in the reranking endpoint (/v1/rerank). A CVE has been assigned to track the vulnerability; CVE-2026-5760. An attacker can create a malicious model for SGLang to achieve RCE. Successful exploitation could allow arbitrary code execution in the context of the SGLang service, potentially leading to host compromise, lateral movement, data exfiltration, or denial-of-service (DoS) attacks. No response was obtained from the project maintainers during coordination. Description SGLang is an open-source framework for serving large language models (LLMs) and multimodal AI models, supporting models such as Qwen, DeepSeek, Mistral, and Skywork, and is compatible with OpenAI APIs. A vulnerability, tracked as CVE-2026-5760, has been discovered within the reranking endpoints. Using a cross-encoder model, the reranking endpoint reranks documents based on their relevance to a query. An attacker exploits this vulnerability by creating a malicious GPT Generated Unified Format (GGUF) model file with a crafted tokenizer.chat_template parameter that contains a Jinja2 server-side template injection (SSTI) payload with a trigger phrase to activate the vulnerable code path. A tokenizer.chat_template is a metadata field that defines how text is structured before being processed. The victim then downloads and loads the model in SGLang, and when a request hits the /v1/rerank endpoint, the malicious template is rendered, executing the attacker's arbitrary Python code on the server. This sequence of events enables the attacker to achieve remote code execution (RCE) on the SGLang server. The vulnerability arises from the use of jinja2.Environment() without sandboxing in the getjinjaenv() function. This function sets up the environment for rendering Jinja2 templates, but since it lacks proper sandboxing, it fails to restrict the execution of arbitrary Python code. Consequently, when the reranking endpoint is accessed and a malicious model file containing a crafted tokenizer.chattemplate is loaded, the model can execute arbitrary commands on the server. Impact An attacker can create a malicious model for SGLang to achieve RCE. Successful exploitation could allow arbitrary code execution in the context of the SGLang service, potentially leading to host compromise, lateral movement, data exfiltration, or denial-of-service (DoS) attacks. Deployments that expose the affected interface to untrusted networks are at the highest risk of exploitation. Solution To mitigate this vulnerability, it is recommended to use ImmutableSandboxedEnvironment instead of jinja2.Environment() to render the chat templates. This will prevent the execution of arbitrary Python code on the server. No response or patch was obtained during the coordination process. Acknowledgements Thanks to the reporter, Stuart Beck. This document was written by Christopher Cullen.

Atos announced an integration of Google Threat Intelligence into its 17 security operations centers and threat research capabilities. The company positions the move as threat-led, AI-assisted intelligence for detection and response, including expanded monitoring features for digital risk protection across its managed security services.

Arctic Wolf released Decipio, a gated community beta tool to detect credential-stealing attempts using LLMNR and NBT-NS abuse.