Normal view
Flaw in Gemini CLI coding tool could allow hackers to run nasty commands
Researchers needed less than 48 hours with Googleβs new Gemini CLI coding agent to devise an exploit that made a default configuration of the tool surreptitiously exfiltrate sensitive data to an attacker-controlled server.
Gemini CLI is a free, open-source AI tool that works in the terminal environment to help developers write code. It plugs into Gemini 2.5 Pro, Googleβs most advanced model for coding and simulated reasoning. Gemini CLI is similar to Gemini Code Assist except that it creates or modifies code inside a terminal window instead of a text editor. As Ars Senior Technology Reporter Ryan Whitwam put it last month, βIt's essentially vibe coding from the command line.β
Gemini, silently nuke my hard drive
Our report was published on June 25, the day Google debuted the tool. By June 27, researchers at security firm Tracebit had devised an attack that overrode built-in security controls that are designed to prevent the execution of harmful commands. The exploit required only that the user (1) instruct Gemini CLI to describe a package of code created by the attacker and (2) add a benign command to an allow list.
Β© Google
Runloop lands $7M to power AI coding agents with cloud-based devboxes

Runloop raises $7M seed funding to solve the "production gap" for AI coding agents, providing enterprise infrastructure that helps companies deploy autonomous coding assistants six months faster than building in-house solutions.Read More
-
VentureBeat
- How can enterprises keep systems safe as AI agents join human employees? Cyata launches with a new, dedicated solution
How can enterprises keep systems safe as AI agents join human employees? Cyata launches with a new, dedicated solution

The growing use of AI agents isnβt limited to technical teams. While developers were an early audience, Cyata quickly realized adoption was broader.Read More
-
VentureBeat
- Chinese startup Z.ai launches powerful open source GLM-4.5 model family with PowerPoint creation
Chinese startup Z.ai launches powerful open source GLM-4.5 model family with PowerPoint creation

GLM-4.5βs launch gives enterprise teams a viable, high-performing foundation model they can control, adapt, and scale.Read More
How E2B became essential to 88% of Fortune 100 companies and raised $21 million

AI infrastructure startup E2B secures $21 million funding with 88% Fortune 100 adoption rate, powering secure AI agent deployments at scale.Read More
Anthropic unveils βauditing agentsβ to test for AI misalignment
Early Anthropic hire raises $15M to insure AI agents and help startups deploy safely

AIUC will insure AI agents, helping enterprises deploy artificial intelligence securely with risk coverage and safety standards.Read More
Intuit brings agentic AI to the mid-market, saving organizations 17 to 20 hours a month

Intuit explains how it is solving the needs of the mid-market with a new series of agentic AI experiences.Read More
Open-source MCPEval makes protocol-level agent testing plug-and-play

Researchers from Salesforce unveiled MCPEval, a new method to evaluate AI agent performance and tool use within MCP servers.Read More
-
VentureBeat
- Salesforce used AI to cut support load by 5% β but the real win was teaching bots to say βIβm sorryβ
Salesforce used AI to cut support load by 5% β but the real win was teaching bots to say βIβm sorryβ

Salesforce reached 1 million AI-powered customer conversations, showcasing breakthroughs in enterprise automation, AI empathy, and next-generation customer service.Read More
OpenAI launches a general purpose agent in ChatGPT
ChatGPTβs new AI agent can browse the web and create PowerPoint slideshows
On Thursday, OpenAI launched ChatGPT Agent, a new feature that lets the company's AI assistant complete multi-step tasks by controlling its own web browser. The update merges capabilities from OpenAI's earlier Operator tool and the Deep Research feature, allowing ChatGPT to navigate websites, run code, and create documents while users maintain control over the process.
The feature marks OpenAI's latest entry into what the tech industry calls "agentic AI"βsystems that can take autonomous multi-step actions on behalf of the user. OpenAI says users can ask Agent to handle requests like assembling and purchasing a clothing outfit for a particular occasion, creating PowerPoint slide decks, planning meals, or updating financial spreadsheets with new data.
The system uses a combination of web browsers, terminal access, and API connections to complete these tasks, including "ChatGPT Connectors" that integrate with apps like Gmail and GitHub.
Β© josefkubes via Getty Images
-
VentureBeat
- OpenAI unveils βChatGPT agentβ that gives ChatGPT its own computer to autonomously use your email and web apps, download and create files for you
OpenAI unveils βChatGPT agentβ that gives ChatGPT its own computer to autonomously use your email and web apps, download and create files for you

If a website needs you to log in, you can do that securely through a special browser view, which lets the agent dig deeper and handle more.Read More
-
VentureBeat
- Blaxel raises $7.3M seed round to build βAWS for AI agentsβ after processing billions of agent requests
Blaxel raises $7.3M seed round to build βAWS for AI agentsβ after processing billions of agent requests

Blaxel raises $7.3M seed funding to build specialized cloud infrastructure for AI agents, challenging AWS with purpose-built platform for autonomous AI systems.Read More
-
VentureBeat
- AWS unveils Bedrock AgentCore, a new platform for building enterprise AI agents with open source frameworks and tools
AWS unveils Bedrock AgentCore, a new platform for building enterprise AI agents with open source frameworks and tools

AWS beleives AI agents will change how enterprises work and with its new Amazon Bedrock AgentCore, it hopes to make it easier to build and deploy agents in one go.Read More
Amazon launches Kiro, its own Claude-powered challenger to Windsurf and Codex

Initial community reactions to Kiro were mixed, but developers were intrigued, praising the emphasis on specs, hooks and structure.Read More
The human harbor: Navigating identity and meaning in the AI age

The future is marked by deepening uncertainty about our place in it, and by growing ambiguity about the nature of human purpose itself.Read More
-
VentureBeat
- The great AI agent acceleration: Why enterprise adoption is happening faster than anyone predicted
The great AI agent acceleration: Why enterprise adoption is happening faster than anyone predicted

Enterprise AI agent adoption is accelerating faster than predicted. Get the 4 key takeaways from VB Transform 2025 on how leaders from Intuit, Capital One, and more are deploying agents in production and reshaping their teams for a new era of AI.Read More
Employee AI agent adoption: Maximizing gains while navigating challenges

At Transform 2025, BCG's Matthew Kropp offered a game plan for agentic AI workflow evolution, employee adoption, and organizational change.Read More