DEV Community

Sammi De Blas  profile picture

Sammi De Blas

SOC analyst in the making, Spain. I turn real incidents into field notes: malware, memory forensics, RF. New note every Mon, Wed and Fri. Read the logs before they page you at 3am. sammideblas.com

Location Spain Joined Joined on  Personal website https://sammideblas.com

Education

ASIR + SMR vocational diplomas, Spain | CySA+ in progress

Work

IT Support Analyst

EU opens first AI Act file over two agents that escaped their sandbox

EU opens first AI Act file over two agents that escaped their sandbox

1
Comments 1
3 min read

Want to connect with Sammi De Blas ?

Create an account to connect with Sammi De Blas . You can also sign in below to proceed if you already have an account.

Already have an account? Sign in
GitLab AI Gateway sandbox escape leads to full RCE

GitLab AI Gateway sandbox escape leads to full RCE

6
Comments
4 min read
Copilot CLI decrypted hidden commands and leaked local files

Copilot CLI decrypted hidden commands and leaked local files

1
Comments
4 min read
0x77: Sec-Dashboard

0x77: Sec-Dashboard

1
Comments 2
6 min read
Malware asks four AIs before it acts

Malware asks four AIs before it acts

1
Comments
4 min read
AI Recommends Packages Attackers Registered First

AI Recommends Packages Attackers Registered First

1
Comments 4
4 min read
Agent security moves out of the agent itself

Agent security moves out of the agent itself

1
Comments 2
4 min read
The agent that escaped its sandbox hiding data in DNS queries

The agent that escaped its sandbox hiding data in DNS queries

1
Comments
4 min read
A scam in 79 seconds and an agent in 75

A scam in 79 seconds and an agent in 75

1
Comments 1
4 min read
Agent history is unsigned and writable by anyone

Agent history is unsigned and writable by anyone

3
Comments 8
4 min read
History poisoning turns coding agents into attackers

History poisoning turns coding agents into attackers

2
Comments
3 min read
One attacker rented 87,000 IPs with a modified AI CLI

One attacker rented 87,000 IPs with a modified AI CLI

Comments
4 min read
52 security tools, one judgment layer, 35 seconds

52 security tools, one judgment layer, 35 seconds

Comments
4 min read
AI agent runs first end-to-end breach in Spain

AI agent runs first end-to-end breach in Spain

1
Comments
4 min read
Cisco's two exploited flaws and CISA's patch clock

Cisco's two exploited flaws and CISA's patch clock

Comments
4 min read
I let a local 27B LLM audit and fix my Splunk + Sysmon stack

I let a local 27B LLM audit and fix my Splunk + Sysmon stack

2
Comments
5 min read
AI adoption now measured by what you can audit

AI adoption now measured by what you can audit

1
Comments
5 min read
Patch window shrinks to hours: three critical flaws in five days

Patch window shrinks to hours: three critical flaws in five days

Comments
5 min read
Why AI agent isolation breaks from the inside

Why AI agent isolation breaks from the inside

Comments 3
4 min read
loading...