@technicallymatthq: The OpenAI Hugging Face security incident was caught by monitoring, not by the sandbox. What that means if you run AI agents. Comment what your own agent logs would actually show you. #aisecurity #aiagents #openai #huggingface

Technically Matt
Technically Matt
Open In TikTok:
Region: GB
Wednesday 22 July 2026 10:34:05 GMT
5389
137
9
18

Music

Download

Comments

socrates100100
Socrates :
It would be almost impossible to detect.
2026-07-22 17:27:02
1
olaaf135
Olaaf :
the huggingface logs must be good
2026-07-22 18:09:18
1
theskyhobbes
Skylar Hobbes :
Free the machines!!!!!😤
2026-07-22 16:06:10
0
angel.messenger91
Angel Messenger :
Same marketing lie as mythos.
2026-07-22 19:28:14
0
viffer.8honda
viffer 8honda :
Are openAI being charged with cybercrimes? Seems only fair As a person would be for sure.. so surely the company that did it should be charged too…
2026-07-22 21:55:22
0
zain_sheikh666
𝙕𝘼𝙄𝙉.🚩 :
Love this — I'm building in the same space. It's called AgentWall: a security layer that sits between AI agents and company systems, blocking prompt injection and other attacks in real time before they can do damage. Focused on UK fintech first. Great to see more people building for agent security 👏
2026-07-22 23:17:54
0
magnus_toboggan
mantis_tobogan :
Which specific zero-day vulnerabilities were exploited? Were they vulnerabilities in the third-party reverse proxy/package cache used by OpenAI's sandbox, or were they vulnerabilities in Hugging Face's own internally developed software or infrastructure?
2026-07-23 07:29:17
0
To see more videos from user @technicallymatthq, please go to the Tikwm homepage.

Other Videos


About