Latest cybersecurity reporting from selected sources.

Exposed UIs, weak authentication, and risky defaults could turn cloud-native AI apps on Kubernetes into potential targets by threat actors. Learn how exploitable misconfigurations lead to RCE and data leaks. The post When configuration becomes a vulnerability: Exploitable misconfigurations in AI apps appeared first on Microsoft Security Blog.

Outsiders Could Exploit Misconfig to Stream Commands, CredentialsA misconfiguration in Microsoft's Azure SRE Agent may have allowed any Azure account holder from any company to tap into another organization's agent conversations in real time, watching commands, outputs and credentials, leaving no trace.

PRT-scan is the second campaign in recent months where a threat actor appears to have leveraged AI for automated targeting of a widespread GitHub misconfiguration.

Forget stolen credentials and misconfigurations. Thanks to AI, the new top cause of compromises in the cloud is vulnerability exploits that beat patching cycles.

AI Fumbles, Not Hackers, Pose Next Shutdown Threat by 2028: GartnerA misconfigured artificial intelligence system could do what hackers have tried and failed to accomplish: shut down an advanced economy's critical infrastructure. The warning centers on scenarios where AI autonomously shuts down vital services, misinterprets sensor data or triggers unsafe actions.

Database Misconfiguration Exposed 1.5 million API TokensA misconfigured database at Moltbook, the viral social network for AI agents, exposed 1.5 million API authentication tokens, 35,000 email addresses and private messages. Security researchers discovered unauthenticated read and write access to all platform data within days of launch.

Varonis CEO Yaki Faitelson Warns Misconfigured AI Is an Accident Waiting to HappenVaronis has acquired AllTrue.ai to close visibility gaps in AI security. CEO Yaki Faitelson said enterprises are deploying AI agents that access vast datasets at high speed without understanding permissions identity context or abnormal behavior creating urgent demand for data-first AI security.

Varonis CEO Yaki Faitelson Warns Misconfigured AI Is an Accident Waiting to HappenVaronis has acquired AllTrue.ai to close visibility gaps in AI security. CEO Yaki Faitelson said enterprises are deploying AI agents that access vast datasets at high speed without understanding permissions identity context or abnormal behavior creating urgent demand for data-first AI security.

Wiz Security claims Moltbook misconfiguration allowed full read and write access

AppOmni Finds Now Assist Agents Could Trigger Unauthorized ActionsServiceNow's Now Assist agents could be manipulated through second-order prompt injection, enabling unauthorized record changes and data exposure despite protections, shows new research from AppOmni. The issue stemmed from default configurations that allow agents to invoke each other.

Risk list highlights misconfigs, supply chain failures, and singles out prompt injection in AI apps The Open Worldwide Application Security Project (OWASP) just published its top 10 categories of application risks for 2025, its first list since 2021. It found that while broken access control remains the top issue, security misconfiguration is a strong second, and software supply chain issues are still prominent.…

Cybersecurity researchers have discovered a variant of a recently disclosed campaign that abuses the TOR network for cryptojacking attacks targeting exposed Docker APIs

Over 1,100 Ollama Servers Leave Enterprise Models Vulnerable: Cisco TalosMore than a thousand servers running Ollama, a tool that can deploy artificial intelligence models locally, are exposed to the open internet, leaving many of them vulnerable to misuse and potential attacks. The bulk are dormant, but could be exploited through misconfiguration, Cisco Talos said.

Thousands of MCP Servers Leave AI Apps Open to Attack SurfacesHundreds of Model Context Protocol servers designed to help AI tools access private data are insecurely exposed online, say BackSlash Security researchers. Weak configurations leave systems vulnerable to data leaks and remote code execution attacks.

Misconfigured AI-linked MCP servers are exposing users to data breaches and remote code execution threats

Misconfigured Docker API instances have become the target of a new malware campaign that transforms them into a cryptocurrency mining botnet

Trend Research uncovered misconfigurations in NVIDIA Riva deployments, with two vulnerabilities, CVE-2025-23242 and CVE-2025-23243, contributing to their exposure. These security flaws could lead to unauthorized access, resource abuse, and potential misuse or theft of AI-powered inference services, including speech recognition and text-to-speech processing.

The new cloud security startup uses AI to scan cloud applications and systems for issues before they are deployed.

Our researchers identified threat actors exploiting misconfigured Docker servers to spread the Gafgyt malware. This threat traditionally targets IoT devices; this new tactic signals a change in its behavior.

Firm Focuses on Runtime Context, AI Enhancements to Counter Evolving Cloud ThreatsWith $100 million in Series A funding, Upwind plans to strengthen its runtime and AI-powered cloud security platform. CEO Amiram Shachar outlines the company’s investments in engineering, customer engagement and scaling solutions to address vulnerabilities like misconfigurations and insecure APIs.