AI Under Watch: Security, Regulation, and the Unexpected!

Hey Devs,

Let’s be honest—AI news is starting to feel like a rerun. Same tools, same headlines, and way too many “AI is taking your job” posts. But under all that noise, weird and crazy things are still happening.

This week, we’ve got jailbreaks, strange bugs, and AI doing stuff no one expected (again). We’re also looking at some cool tools and one security story that’ll make you double-check your passwords.

Let’s dive in.

🧠 AI & Regulation

EU AI Act Phase 1 Enforcement

• February 2, 2025 saw the ban on “unacceptable risk” AI (e.g., subliminal manipulation, social scoring) and new AI‑literacy requirements for providers and deployers European Parliament

• The European Commission simultaneously published guidelines on prohibited practices—clarifying rules around emotion recognition, bio-metric profiling, and general‑purpose AI transparency Global Law Firm | Paul Hastings LLP

• Despite U.S. political pushback, the EU reaffirmed its commitment to enforce high standards, even as code‑of‑practice talks aim to conclude in April 2025.

🚨 Cyber-Security Watch

DaVita Ransomware

• On April 14, 2025, DaVita Inc. reported a ransomware incident that encrypted elements of its network; the company isolated affected systems, continued patient care, and engaged law enforcement and external experts.

Hertz Vendor Breach

• Also on April 14, 2025, Hertz disclosed that hackers exploited zero‑day flaws in its vendor Cleo Communications’ file‑transfer platform (Oct–Dec 2024), potentially exposing customer contact, payment, and ID data—its own network remained intact.

SVG Phishing Explosion

• Trustwave SpiderLabs reports an 1,800 % jump in phishing campaigns using SVG attachments in early 2025, driven by PhaaS platforms like Tycoon2FA and Mamba2FA, which embed malicious scripts in XML‑based images to bypass email scanners.

🔧 Tech Spotlight: Microsoft Security Copilot Agents

Phishing Triage Agent

• Introduced at Microsoft Secure 2025, this agent uses LLM‑driven semantic analysis to automatically resolve 95 % of user‑reported phishing incidents, providing natural‑language explanations and learning from analyst feedback Microsoft Tech Community

Previewing 11 AI Security Agents

• In April 2025, Microsoft will preview agents for Alert Triage, Conditional Access Optimization, Vulnerability Remediation, Threat Intelligence Briefing, Insider Risk, Data Loss Prevention, and more—each designed to triage, prioritize, and automate SOC workflows under a Zero Trust model Microsoft

😂 Meme Lab & ⚡ Lightning Tip

🔧 Lightning Tip:
Scan your network for exposed management ports (SSH, RDP) in one command:

nmap -p 22,3389 --open 192.168.1.0/24

This command uses nmap, short for Network Mapper, a powerful open-source tool built in the Bash/shell scripting environment (used on Linux, Mac, and Windows via tools like PowerShell or WSL). It's written in C and Lua, but you don’t need to know those to use it. You just run it through your command-line terminal.

This command scans your local subnet (192.168.1.0/24)—usually your home or office Wi-Fi network—for devices that have SSH (port 22) or RDP (port 3389) open. These are common remote access services used for logging into systems. If they're open and unused, they can become backdoors for attackers. The --open flag filters the results to show only systems with those ports currently accessible.

🎯 Hot Take

Wrapping up…
From EU compliance milestones to high‑profile cyberattacks and AI‑powered defense agents, this issue keeps you ahead of the curve. Until next Sunday—stay alert, stay updated.

P.S. Got questions or feedback? Just shoot us an email at [email protected]!

Want to make money online without launching a sketchy crypto token? Start a newsletter on beehiiv. It powers this one, and it’s super easy — even for sleep-deprived devs and caffeine-powered students.

Partner Disclosure: Please note that some of the links in this post are affiliate links, which means if you click on them and make a purchase, I may receive a small commission at no extra cost to you. This helps support my work and allows me to continue to provide valuable content. I only recommend products that I use and love. Thank you for your support!