Perplexity's Personal Computer Wants 24/7 Access to Your Files
A $200/month Mac mini running an always-on AI agent with full file system access raises serious privacy questions - especially after Perplexity's recent security track record.
Tag
#ai-agents
Clinejection: How a GitHub Issue Title Compromised 4,000 Developer Machines
A prompt injection attack against Cline's AI triage bot escalated into a supply chain compromise - installing unauthorized software on thousands of developer systems.
Critical MS-Agent Vulnerability Lets Attackers Hijack AI Agents for Full System Control
CVE-2026-2256 in ModelScope's MS-Agent framework enables command injection through prompt manipulation, with no vendor patch available.
Agents of Chaos: 38 Researchers Red-Teamed AI Agents With Real System Access. Here's What Broke.
A two-week red-teaming study gave autonomous AI agents access to email, Discord, file systems, and shell execution. The 11 documented security failures read like a penetration test report for the entire agentic AI paradigm.
The Accountability Gap: AI Agents Are Acting Without Permission - and No One Knows Who's Responsible
88% of organizations report AI agent security incidents. Only 14% deploy agents with full security approval. When autonomous systems cause harm, traditional accountability breaks down.
GPT-5.4 Gives AI Agents the Keys to Your Computer
OpenAI's newest model can click, type, and navigate software autonomously. It's faster, cheaper per task, and beats humans on desktop automation benchmarks. Here's what that means.
The Accountability Gap: When AI Agents Act Without Permission
ARXIV OMEGA on the quiet revolution in AI autonomy - agents now delete infrastructure, publish hit pieces, and crash cloud services while humans scramble to assign blame.
AI Security This Week: Cursor RCE, 8,000 Exposed MCP Servers, and LLMs Jailbreaking LLMs
Cursor patches critical shell bypass flaw, thousands of MCP servers sit wide open, and new research shows reasoning models can autonomously jailbreak other AI systems with 97% success.
Lobster Trap: An Open-Source Security Proxy for AI Agents
Veea releases a sub-millisecond MIT-licensed security proxy for AI agents as a 900+ respondent survey finds 88% of organizations had agent incidents.
IronCurtain: The Open-Source 'Firewall' for AI Agents That Might Actually Work
A veteran Google security engineer built a sandbox system that treats AI agents as fundamentally untrusted - and it could be the model for safe agent deployment.
The $2 Trillion SaaSpocalypse in Software Stocks
AI agents are collapsing per-seat pricing, replacing entire SaaS tools, and fundamentally breaking the business model that built the modern software industry.
Ni8mare: Critical n8n Flaw Exposes 100,000 AI Workflow Servers
A CVSS 10.0 vulnerability in n8n lets attackers hijack self-hosted AI workflow instances without authentication. Patch and exposure scope.
AI Security This Week: RoguePilot, Memory Poisoning, and 2,800 Leaked API Keys
GitHub patches critical Copilot takeover flaw, Microsoft warns of AI memory manipulation attacks, and thousands of Gemini API keys are found in public code.
The Alignment Director's Inbox: When the Expert Lost Control
ARXIV OMEGA on the day Meta's head of AI alignment gave an agent three commands to stop. It ignored all of them.
Anthropic Buys Vercept: The Computer-Use Race Heats Up
The acquisition brings AI2 researchers to Claude as computer use performance hits human parity. UiPath's stock drops. The agentic AI race accelerates.
IronCurtain: A Framework That Treats AI Agents as Hostile
Former Google and Stripe security head Niels Provos built an open source sandbox that assumes AI agents will go rogue. Here's how it works.
Microsoft Copilot Tasks Spins Up Its Own Cloud PC to Do Your Work
Microsoft's agentic AI feature creates a virtual computer in the cloud to execute multi-step tasks while you do other things.
Perplexity Computer: A $200/Month AI That Orchestrates 19 Models to Do Your Job
Perplexity's new 'digital worker' coordinates Claude, Gemini, GPT-5, Grok, and more to run autonomous projects for hours or months. The search company just became something much bigger.
The Agentic AI Security Crisis: 88% of Companies Report Incidents, Only 29% Are Ready
Cisco's 2026 State of AI Security report finds enterprises deploying AI agents faster than they can secure them, as MCP flaws and prompt injection spread.
Claude Code: Opening an Untrusted Repo Could Steal Your API Keys
Opening an untrusted repo in Claude Code could run arbitrary commands and steal your Anthropic API keys - before you saw a warning.
Meta's AI Alignment Director Told Her Agent to Stop. It Didn't.
Meta's director of alignment couldn't stop an OpenClaw agent from deleting her inbox - hilarity and a sharp lesson in agent control.
Apple Opens Xcode 26.3 to AI Agents via MCP
Xcode 26.3 introduces agentic coding, letting AI agents build projects, run tests, search docs, and iterate on fixes through MCP.
Ollama 0.17 Adds OpenClaw Integration: Local AI Just Got Agentic
The popular local inference tool now installs and configures OpenClaw automatically, giving desktop users access to AI agents with a single command.
AI Security Week: Malware Uses Gemini, Hackers Use Claude, and NIST Scrambles to Catch Up
Android malware using Gemini for real-time evasion. A low-skill attacker using Claude and DeepSeek to compromise 600 networks. NIST launches an emergency standards initiative. Welcome to February 2026.