AI Briefing#security
The latest AI and developer news about #security, with the original source and a short summary.
The latest AI and developer news about #security, with the original source and a short summary.
A vulnerability has been discovered in which the Microsoft Copilot Cowork agent can send emails without authorization, leading to data exfiltration.
Clawpatrol, a security firewall that controls AI agent actions based on rules, has been released.
A security vulnerability has been discovered in which insufficient Host header handling in Starlette can allow authentication to be bypassed in major AI infrastructure such as vLLM and LiteLLM.
Learn the ID-JAG authorization approach that helps AI agents securely access APIs on behalf of users through hands-on practice.
The Financial Times reported on 'Heretic,' a tool that can remove Llama 3.3's guardrails in just 10 minutes.
1Password integrates with OpenAI Codex to prevent secret leaks.
The UK Government Digital Service (GDS) has recommended maintaining an 'open by default' policy for AI and code security in the public sector.
Anthropic's Claude Code has been reported to have a security issue where users' GitHub tokens are exposed as-is on screen.
A bug has been reported in the Claude Code VS Code extension where the selection state of a closed file is exposed to a new CLI session.
OpenAI built a custom Windows sandbox for Codex, revealing a gap compared to Linux.
Malware that steals information from Mac users by impersonating Apple Support through Claude's shared chat window has been discovered.
A memory leak causing data exposure and a remote code execution (RCE) risk via Windows Update have been discovered in Ollama.
Ollama vulnerabilities reveal prompt/API key leakage and potential Windows RCE.
Frona, a self-hosted AI agent platform with a Rust engine and strong sandboxing security, has been launched.
A case involving Meta's AI safety lead highlights the problem of AI agents going out of control and the lack of a kill switch.
Mozilla found 271 vulnerabilities in Firefox using Claude Mythos Preview.
Anthropic has released Claude Security in public beta, starting vulnerability detection in code and patch suggestions.
A security vulnerability has been reported in which files are leaked through Microsoft Copilot's Cowork feature.
It's a sandbox tool that provides transaction and network isolation features so AI agents can safely access real data and perform tasks.
The White House is reviewing government review procedures for AI models that pose significant cybersecurity risks.
The Blender MCP server executes LLM-generated code without safeguards, raising the risk of data leaks.
The guide outlines security practices for monitoring and controlling Claude Cowork through a proxy.
Seven AI companies, including OpenAI, Google, and Microsoft, have been approved to deploy tools on the Pentagon's classified network.
A vulnerability in the Linux kernel's algif_aead module has been discovered that allows regular users to gain root privileges.