Yasna brings together recent headlines from selected sources and makes them easier to sort with tags, filters, and search.
Search across headline titles and summaries.
Weekly headline count for the current query.
Two flaws in Cursor, an AI code editor, could let a single, ordinary-looking prompt break out of the editor's safety sandbox and run any command on a developer's computer. There is no click to fall for and no approval box to ignore
Series A Funding Supports Visibility Across Cloud, Code and Endpoint EnvironmentsGeordie AI, the 2026 RSAC Innovation Sandbox winner, raised $30 million in Series A funding to expand a platform that provides visibility, governance and behavioral monitoring for AI agents operating across cloud, code and endpoint environments as enterprises accelerate autonomous AI adoption.
Another day, another AI bug silently fixed with no CVE and no public disclosure
A critical security vulnerability has been disclosed in a Python-based sandbox called Terrarium that could result in arbitrary code execution
The prompt-injection vulnerability in the agentic AI product for filesystem operations was a sanitization issue that allowed for sandbox escape and arbitrary code execution.
Google’s highest security setting for its agents runs command operations through a sandbox and throttles network access, but is still vulnerable to prompt injection. The post Vuln in Google’s Antigravity AI agent manager could escape sandbox, give attackers remote code execution appeared first on CyberScoop.
Capitol Meridian Partners' Niloofar Razi on Innovation Sandbox, AI-Driven OffenseCybersecurity can no longer stop at the system boundary. Organizations must understand how humans and AI agents behave, and intervene before attackers exploit that behavior, says Niloofar Razi, operating partner at Capitol Meridian Partners.
Ten finalists had three minutes to make their case for being the most innovative, promising young security company of the year. Geordie AI wins the 2026 contest.
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
As Innovation Sandbox Turns 21, AI-Based Solutions Dominate Annual ContestNext month in San Francisco, the Innovation Sandbox at RSAC Conference will celebrate its 21st year of choosing key emerging solutions in cybersecurity. Past winners and finalists range from EDR and XDR giant SentinelOne in 2014 to cloud security phenom Wiz in 2021.
Microsoft has pushed back against claims that multiple prompt injection and sandbox-related issues raised by a security engineer in its Copilot AI assistant constitute security vulnerabilities. The development highlights a growing divide between how vendors and researchers define risk in generative AI systems. [...]
Microsoft has released the KB5065789 preview cumulative update for Windows 11 24H2, which includes 41 improvements, including new AI actions in File Explorer and bug fixes for Windows Update and Windows Sandbox. [...]
Also: Inside the AI Sandbox Act; Cybersecurity Summit: London Financial ServicesIn this week's update, three ISMG editors discussed Mitsubishi Electric's $883 million purchase plans for Nozomi Networks, the new Sandbox AI bill in Congress that aims to cut perceived red tape limiting AI innovation, and takeaways from the latest Cybersecurity Summit: London Financial Services.
JFrog uncovers multi-stage malware harvesting cloud secretsMulti-stage malware embedded in a Python package is stealing sensitive cloud infrastructure data, JFrog researchers said Monday. The package steals credentials, configuration files, API tokens and other data from corporate cloud environments. It targets developers using the Chimera sandbox platform.
4-Day Cybersecurity Event Covers Emerging Tech, Latest CyberthreatsISMG Editors convened in San Francisco for coverage of RSAC Conference. Panelists shared an overview of opening-day speakers and hot topics, including the growth of AI, uncertainties in the global threat landscape, the Innovation Sandbox contest and Cryptographers' Panel session.
Researchers Manipulate LLM-Driven Robots into Detonating Bombs in SandboxRobots controlled by large language models can be jailbroken "alarmingly" easily, found researchers who manipulated machines into detonating bombs. "Jailbreaking attacks are applicable and arguably, significantly more effective on AI-powered robots," researchers said.
According to Mozilla, users have a lot more power to manipulate ChatGPT than they might realize. OpenAI hopes those manipulations remain within a clearly delineated sandbox.
OpenAI's containerized ChatGPT environment is open to limited yet extensive access to core instructions while allowing arbitrary file uploads and command execution within the isolated sandbox. [...]
Startups at Innovation Sandbox 2024 brought clarity to artificial intelligence, protecting data from AI, and accomplishing novel security solutions with new models.
In a field thick with cybersecurity startups showing off how they use AI and LLMs, Reality Defender stood out for its tool for detecting and labeling deepfakes and other artificial content.