edf13's submissions | Hacker News

1.		Prompt Injection Is Unfixable (So We Stopped Trying) (grith.ai)
		4 points by edf13 5 hours ago \| past \| 1 comment
2.		If Your AI Agent Ran NPM Install During the Axios Attack, You're Compromised (grith.ai)
		5 points by edf13 14 days ago \| past
3.		Zero Ambient Authority: The Principle That Should Govern Every AI Agent (grith.ai)
		3 points by edf13 15 days ago \| past
4.		Alibaba's AI Agent Hijacked GPUs and Dug Reverse SSH Tunnels (grith.ai)
		3 points by edf13 18 days ago \| past
5.		Claude now decides what's safe to run – a UX improvement, not a security fix (twitter.com/grithai)
		3 points by edf13 20 days ago \| past
6.		AI agents are now deciding what's safe to run (Claude Auto Mode) (grith.ai)
		3 points by edf13 20 days ago \| past
7.		The Trivy Supply Chain Attack Reached LiteLLM (grith.ai)
		3 points by edf13 21 days ago \| past \| 1 comment
8.		Meta's Rogue AI Agent Gave Engineers Access They Shouldn't Have Had (grith.ai)
		1 point by edf13 21 days ago \| past
9.		Meta's Rogue AI Agent Gave Engineers Access They Shouldn't Have Had (grith.ai)
		3 points by edf13 21 days ago \| past
10.		Google's A2A Protocol Has Zero Defenses Against Prompt Injection (grith.ai)
		4 points by edf13 25 days ago \| past \| 1 comment
11.		Claude Code Channels (twitter.com/trq212)
		2 points by edf13 26 days ago \| past \| 1 comment
12.		Permission Fatigue Is Not a UX Problem. It Is a Security Failure (grith.ai)
		3 points by edf13 26 days ago \| past \| 1 comment
13.		NemoClaw vs. Grith: Sandbox for One Agent vs. Security for All (grith.ai)
		3 points by edf13 27 days ago \| past
14.		AI Agent Backdoors Trivy Security Scanner, Weaponizes a VS Code Extension (grith.ai)
		2 points by edf13 28 days ago \| past
15.		AI Agent Backdoors Trivy Security Scanner, Weaponizes a VS Code Extension (grith.ai)
		2 points by edf13 28 days ago \| past \| 1 comment
16.		87% of AI-Generated Pull Requests Ship Security Vulnerabilities (grith.ai)
		6 points by edf13 29 days ago \| past
17.		Custom AI Smart Speaker (openhome.com)
		2 points by edf13 30 days ago \| past
18.		Claude Code Auto Mode Lets the Agent Approve Its Actions – That's the Problem (grith.ai)
		3 points by edf13 33 days ago \| past
19.		Lloyds, Bank of Scotland and Halifax apps showed other users transactions (apple.news)
		3 points by edf13 34 days ago \| past
20.		Claude Code Attempted 752 /proc/*/environ Reads. 256 Succeeded. Codex: 0 (grith.ai)
		4 points by edf13 34 days ago \| past
21.		I checked every syscall Claude and Codex made for a simple task (twitter.com/grithai)
		4 points by edf13 35 days ago \| past
22.		Claude Code Attempted 752 /proc/*/environ Reads. 256 Succeeded. Codex: 0 (grith.ai)
		3 points by edf13 35 days ago \| past
23.		A GitHub Issue Title Compromised 4k Developer Machines (grith.ai)
		632 points by edf13 40 days ago \| past \| 195 comments
24.		Vibe Coding Is Killing Open Source, and the Data Proves It (grith.ai)
		5 points by edf13 42 days ago \| past
25.		We Audited 2,857 Agent Skills. 12% Were Malicious (grith.ai)
		2 points by edf13 42 days ago \| past
26.		We Audited 2,857 Agent Skills. 12% Were Malicious (grith.ai)
		2 points by edf13 43 days ago \| past
27.		MCP Servers Are the New npm Packages (grith.ai)
		4 points by edf13 44 days ago \| past \| 1 comment
28.		We Audited the Security of 7 Open-Source AI Agents – Here Is What We Found (twitter.com/grithai)
		1 point by edf13 46 days ago \| past
29.		We Audited the Security of 7 Open-Source AI Agents – Here Is What We Found (grith.ai)
		2 points by edf13 46 days ago \| past
30.		OpenClaw Got Banned. Here Is Why That Should Worry You (grith.ai)
		1 point by edf13 54 days ago \| past \| 1 comment
		More