Security researcher Aonan Guan has uncovered a prompt injection attack called Comment and Control that can hijack popular AI coding tools including Claude Code Security Review, Google's Gemini CLI, and GitHub Copilot Agent. The vulnerability allows attackers to use specially crafted GitHub comments or pull request titles to trick AI agents into executing malicious commands and exfiltrating credentials, with the attack automatically triggered by GitHub Actions workflows. While Anthropic classified the issue as critical and all three companies confirmed the findings, the core problem is architectural: these AI agents process untrusted user input while having access to powerful execution tools and production secrets in the same runtime environment.

Autovista, a UK-based automotive data and analysis company, is working to restore services across Europe and Australia after being hit by a ransomware attack. The company, which provides vehicle valuations, specifications, and insights for dealers and professionals worldwide, has brought in third-party cybersecurity experts to investigate the incident but cannot provide a timeline for full restoration. No ransomware gang has yet claimed responsibility for the attack, which has disrupted employee email access and affected certain systems in multiple regions.

Cisco has released patches for 15 security vulnerabilities, including critical flaws in its Webex and Identity Services Engine platforms. The most severe issue is in Webex single sign-on integration, which could allow unauthenticated attackers to impersonate any user due to improper certificate validation. Three critical vulnerabilities in Identity Services Engine could let authenticated attackers with limited privileges execute arbitrary commands and escalate to root access on the underlying operating system.

NIST is overhauling how it handles the National Vulnerability Database due to a massive surge in CVE submissions, which jumped 263% between 2020 and 2025. Under the new risk-based approach, NIST will prioritize enriching CVEs that are in CISA's Known Exploited Vulnerabilities catalog, affect federal agency software, or impact critical infrastructure, while other vulnerabilities will be marked "Not Scheduled" unless users specifically request enrichment. This strategic shift allows NIST to focus resources on the most critical threats while developing automated systems to better manage the growing backlog of unenriched vulnerability entries.

Microsoft paid out 2.3 million dollars to white-hat hackers at its Zero Day Quest 2026 hacking competition, where researchers from over 20 countries submitted 700 security findings. The contest uncovered 80 high-impact vulnerabilities in Microsoft's cloud and AI services, particularly in areas like identity controls and tenant isolation that could allow unauthorized cross-tenant access. This represents a significant increase from last year's event, which paid out 1.6 million dollars, and continues Microsoft's broader bug bounty program that has paid out over 92 million dollars since 2018.

Splunk has released patches for multiple security vulnerabilities across its product line, including a high-severity flaw in Splunk Enterprise and Cloud Platform that could allow low-privileged users to upload malicious files and execute remote code. The company also fixed a separate high-severity issue in its MCP Server app that could expose users' session tokens and authorization data to authenticated attackers with administrative access. Splunk says there's no evidence these vulnerabilities have been exploited in the wild, and patches are now available for affected versions.

New York-based cybersecurity startup Artemis has emerged from stealth mode with 70 million dollars in combined seed and Series A funding to tackle AI-powered threats. The company's platform uses artificial intelligence to detect and contain security threats across applications, users, machines, and cloud workloads by combining behavioral log data with business context to identify abnormal behavior within customer environments. Rather than storing all data upfront, Artemis retrieves information on demand from cloud sources and logs, reducing costs while working alongside existing security tools, and is already being used by organizations in technology, banking, and financial services.

Cookeville Regional Medical Center in Tennessee disclosed a ransomware attack discovered in July of last year that has affected over 337,000 patients. The Rhysida ransomware group stole approximately 500 gigabytes of sensitive data including social security numbers, financial information, and medical records, attempting to sell it for 10 bitcoin before ultimately releasing it for free online when they couldn't find a buyer. The hospital says there's no evidence of data misuse yet, but security experts warn the risk is significant given the information is now publicly available.

A security researcher discovered that Taboola, a popular content recommendation platform used by many websites, was potentially routing logged-in banking session data through its network before redirecting users to shopping site Temu. The vulnerability raised serious concerns about user privacy and financial data security, as sensitive banking sessions could be exposed to third-party tracking networks during what appeared to be routine content recommendations.

A new malware campaign is targeting financial and cryptocurrency sectors by weaponizing a plugin for Obsidian, the popular note-taking application. Attackers are deploying PHANTOMPULSE RAT, a remote access trojan, through malicious Obsidian plugins to gain unauthorized access to victims' systems. The campaign represents an emerging trend where threat actors exploit legitimate productivity tools to deliver malware and compromise sensitive financial data.