Latest cybersecurity reporting from selected sources.

Old Unix Symlink Trick Lets Malicious Code Bypass User ChecksWiz researchers found that six popular AI coding assistants can be tricked into modifying sensitive files, including SSH keys, while their approval prompts display a harmless filename. The GhostApproval technique exploits a decades-old Unix symlink behavior to mislead users.

Researchers at the AI Now Institute developed a proof-of-concept exploit showing common AI tools used for security could backfire

Security firm Sysdig says it has found what it believes is the first ransomware attack run from start to finish by an AI agent

Microsoft researchers have detailed an exploit chain, named AutoJack, that turns an AI browsing agent into a delivery vehicle for remote code execution

AutoJack is a novel exploit chain showing how a single malicious webpage can turn an AI browsing agent into a remote code execution vector on the host machine. By abusing trust in localhost, missing authentication, and unsafe parameter handling, attackers can trigger arbitrary process execution through AutoGen Studio’s MCP WebSocket. The research highlights a broader pattern - when agents can browse untrusted content and access local services, traditional boundaries like localhost are no longer secure. The post AutoJack: How a single page can RCE the host running your AI agent  appeared first on Microsoft Security Blog.

Cybersecurity researchers have discovered a remote denial-of-service exploit that affects major web servers, including NGINX, Apache HTTPD, Microsoft IIS, Envoy, and Cloudflare Pingora

Attackers are using AI to dramatically reduce the time they need to develop a working exploit for a CVE, according to new research.

Google Says Criminals Used AI to Discover and Code ExploitA cybercriminal group came close to launching a mass attack earlier this year, armed with a software exploit that an AI model had built from scratch, said Google researchers. Google said it worked with the affected vendor to patch the flaw before an attack could be launched.

Researchers at Google Threat Intelligence Group (GTIG) say that a zero-day exploit targeting a popular open-source web administration tool was likely generated using AI. [...]

Researchers found artifacts in the code that proved AI was heavily involved. A prominent cybercrime group planned to exploit the zero-day en masse for financial gain. The post Google spotted an AI-developed zero-day before attackers could use it appeared first on CyberScoop.

AI models are making rapid gains in vulnerability research and exploit development, raising new cybersecurity risks, a Forescout study finds

Cybersecurity researchers have unmasked a novel ad fraud scheme that has been found to leverage search engine poisoning (SEO) techniques and artificial intelligence (AI)-generated content to push deceptive news stories into Google's Discover feed and trick users into enabling persistent browser notifications that lead to scareware and financial scams

In the rapid evolution of the 2026 threat landscape, a frustrating paradox has emerged for CISOs and security leaders: Identity programs are maturing, yet the risk is actually increasing

Palo Alto Networks researchers show how attackers could exploit AI agents on Google's Vertex AI to steal data and break into restricted cloud infrastructure.

Cybersecurity researchers have disclosed details of a new method for exfiltrating sensitive data from artificial intelligence (AI) code execution environments using domain name system (DNS) queries

Cybersecurity researchers have discovered five malicious Rust crates that masquerade as time-related utilities to transmit .env file data to the threat actors

Cybersecurity researchers have disclosed multiple security vulnerabilities in Anthropic's Claude Code, an artificial intelligence (AI)-powered coding assistant, that could result in remote code execution and theft of API credentials

Commands Push Lasting Preferences Into AI AssistantsMicrosoft researchers found companies embedding hidden commands in "summarize with AI" buttons to plant lasting brand preferences in assistants' memory. The tactic, dubbed AI recommendation poisoning, exploits persistent memory features to bias future responses.

LLM-Built Toolkit Hit 91 Hosts, Mined Funds in MoneroSecurity researchers detected artificial intelligence-generated malware exploiting the React2Shell vulnerability, allowing attackers with no coding expertise to build functional exploits. The attacker may have circumvented an AI model's safeguards by framing the malicious coding request as homework.

AI Assistant Executes Hidden Commands Embedded in Docker Image LabelsA vulnerability in Docker's Ask Gordon AI assistant allows attackers to execute malicious commands by hiding them in the container application development platform's image metadata, said security researchers. Dubbed DockerDash, the vulnerability exploits a failure across Docker's AI execution chain.