This may be the worst week in the history of internet security. AI is helping … · @Michaelzsguo · TwKit

@Michaelzsguo

This may be the worst week in the history of internet security. AI is helping people find flaws by the hundreds, and attackers are using AI to exploit them. Anthropic's Jason Clinton says it plainly: the next 18 months may get worse. So do we wait it out, or learn to protect ourselves? If AI helps the bad guys find the holes, it is also the best defence you have. I gave my agent a simple prompt: here's my homelab. Design and build a security system that can defend, monitor, detect and analyze, respond and recover, don't build from scratch and reuse existing open source tools It walked every device and every open port, found a few problems I'd forgotten about, and came back with a detailed plan. Its conclusion: three open-source tools get you 80% of the way. 1. Wazuh (https://github.com/wazuh/wazuh), the watchtower. A small agent on every machine. It pulls all the logs into one place, flags changes to critical files, checks every installed package against CVE databases so you know what to patch, and can block attackers automatically. SIEM, host intrusion detection and vulnerability management in one. 2. CrowdSec (https://github.com/crowdsecurity/crowdsec) , the bouncer. It watches your internet-facing services (SSH, web servers), spots brute-force attempts, scans and exploit probes, and bans the IP. It also uses shared community intelligence: an IP that just attacked someone else gets blocked at your door too. 3. AdGuard Home (https://github.com/AdguardTeam/Adguardhome), the gatekeeper. Every device at home resolves domains through it, and known malware and phishing domains are refused. Your cameras, speakers and printer can't run antivirus, so this is how they get protected. The other 20%: Tailscale access rules, MFA, disk encryption, phone alerts, and backups. You don't need an enterprise product or a budget, just an agent willing to do the work.

曝光 267 · 评论 0 · 点赞 4 · 书签 0