Secure GenAI
Secure GenAI Podcast
Less Regulation, DOGE, 2.7B IoT breach, OpenAI 20M, Nvidia security, Salt Typhoon
0:00
Current time: 0:00 / Total time: -12:21
-12:21

Less Regulation, DOGE, 2.7B IoT breach, OpenAI 20M, Nvidia security, Salt Typhoon

GenAI Safety & Security | Feb 10 - Feb 17, 2025

Notice: FY2024 is coming in Feb 2025.

Highlights

🇺🇸 1. Less AI regulation pushed.

⚠️ 2. DOGE.gov insecure.

🚨 3. IoT breach: 2.7B records.

🕵️ 4. OpenAI: 20M credentials for sale. 

🛡️ 5. NVIDIA: AI cybersecurity platform.

Secure GenAI is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Deep Dive:

🇺🇸 1.Less AI Regulation at Paris Summit [TIME, POLITICO]

  • US & UK didn't sign declaration.

  • US VP Vance advocated opportunity.

  • President Macron pushed pro-business.

  • E.U. regulations received strong criticism.

  • UK: AI safety → AI security institute.

Immediate Mitigation: Be aware that international AI regulatory landscapes are diverging. Consider potential impacts on your org.

⚠️ 2.Anyone Can Push Updates to the DOGE.gov [404 Media]

  • Website insecure.

  • Database editable by anyone.

  • Doge.gov built on Cloudflare Pages.

  • Not hosted on government servers.

  • Pulled from a Cloudflare Pages website.

Mitigation: Ensure that sensitive data is not accessible to unauthorized users.

🚨 3.IoT Data Breach Billions Records [Infosecurity Magazine]

  • Database size: 1.17 Terabytes.

  • Compromised Wi-Fi, passwords and IPs.

  • Linked to Mars Hydro & LG-LED.

  • 57% of IoT devices are vulnerable.

  • Attackers used "nearest neighbor" exploits.

Mitigation: Audit IoT device security, enforce strong passwords, segment networks, and patch vulnerabilities quickly.

🕵️ 4.OpenAI Finds No Evidence of Breach After Hacker Offers to Sell 20 Million Credentials [SecurityWeek, TechRadar]

  • Hacker claimed 20M OpenAI credentials.

  • OpenAI found no compromise.

  • Data likely from infostealer malware.

  • Over 4M bots collected in 2024.

  • Related auth0.openai.com

Immediate Mitigation: Enable MFA, reinforce password security training.

🛡️5.NVIDIA Cybersecurity AI [NVIDIA Blog]

  • Integrates BlueField-3 DPUs & Morpheus AI.

  • Partners: Armis, Check Point, CrowdStrike, Deloitte, WWT.

  • Real-time threat detection for infrastructure.

  • Functions as virtual security overlay.

  • Protects energy, utilities and manufacturing.

Immediate Mitigation: For NVIDIA customers, evaluate and implement the Cybersecurity AI platform for enhanced threat detection in critical infrastructure.

6. (bonus) Salt Typhoon is exploiting Cisco Routers [WIRED]

  • Salt Typhoon still hacking telecoms.

  • Breached five telecoms & ISPs globally.

  • Exploited Cisco routers' web interfaces.

  • Targeted 1000 devices installed.

  • Sanctions don't cause course change.

Immediate Mitigation: Ensure that sensitive data is not accessible with a backup plan.

Thanks for reading Secure GenAI ! This post is public so feel free to share it.

Share

Discussion about this episode