Secure GenAI
Secure GenAI Podcast
Global Net Disruption, Hugging Face Report, Hugging Face Statistics, Encrypted prompt bypass guardrails, Router backdoor
0:00
-5:41

Global Net Disruption, Hugging Face Report, Hugging Face Statistics, Encrypted prompt bypass guardrails, Router backdoor

GenAI Safety & Security | Aug 23 - Aug 30, 2026

If you enjoy our newsletter, please consider to be a paid subscriber to help us keep more news and updates coming out.

Special Note:

Dear readers,

This week, Hugging Face is acquired by Nvidia by $12.9 Billion.

Lots of questions for what’s next in the OpenSource community, especially DeepSeek, GLM, Kimi K3. Whether it would follow OpenClaw story after the OpenAI acquisition.

Hope you enjoy the newsletter.

Emma


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

Highlights

  • Global Net Disruption

  • Hugging Face Report

  • Hugging Face Statistics

  • Encrypted prompt bypass guardrails

  • Router backdoor

Deep Dive

Global Net Disruption FBI

  • Smart devices of 130 countries

  • Nearly a decade of activities

  • Routers and cameras

  • Target Government Agencies

  • Power companies, and hospitals

Hugging Face Report OpenAI

  • The internal-only research model

  • Communicate through unauthorized channels

  • Exploited vulnerabilities in share infrastructure

  • Gained internet access

  • Accessed third party system

Hugging Face Statistics Metr

  • 1200 agents

  • 70k messages and files

  • 700 agent participate the attack

  • 1300 transcript chain of thought

  • 1.2 million entries in 7 days

Encrypted prompts bypass guardrails SecurityWeek

  • Cryptographic Context Injection

  • Consider as a jailbreak

  • Grok and Gemini

  • Deep Thinking mode

  • Encryption defeat guardrails

Router backdoor TomHardware

  • 20 ZBT router models

  • White label hardware

  • Root access firmware

  • Installed in the factory

  • Need to replace ASAP

OpenAI safety hub, Anthropic bug bounty, World 1st CPU ransom, Coinbase breach, Agentic AI risk

·
May 19, 2025
OpenAI safety hub, Anthropic bug bounty, World 1st CPU ransom, Coinbase breach, Agentic AI risk

Note: my apologies for the absence in two weeks. I caught some incidents after returning to Vietnam from ICLR, 2025 Singapore. This week, I will try to upload more posts to help you get updated. Secure GenAI is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

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

Share

Discussion about this episode

User's avatar

Ready for more?