There Are Fewer AI Sandbox Escapes Than You Might Think

Most of the AI hacking incidents you’ve heard of were NOT sandbox escapes, they were all attributable to one misconfiguration at one third party.

Oct 6 • Adrian Sanabria

2

4

1

September 2026

Breach Lessons: the 2022 Uber Hack

Steal my creds once, shame on you, steal them a third time...

Sep 22 • Adrian Sanabria and Vladimir Serov

Chatham House Rule does NOT mean “keep this information private”

Getting this wrong could have dire consequences

Sep 22 • Adrian Sanabria

There’s now a Cyber Weapon Index

Apparently AI models are weapons now and they’re going to kill us all (but that’s a different post)

Sep 16 • Adrian Sanabria

1

1

Breach Lessons: the Capital One Breach

Cloud security lessons for the ages

Sep 15 • Adrian Sanabria and Vladimir Serov

3

1

Breach Lessons: the 2023 British Library Breach

More than just a library - a computer museum that was also powering the library’s infrastructure

Sep 2 • Vladimir Serov and Adrian Sanabria

August 2026

Reviewing initial research on using AI for vulnerability remediation

TL;DR: don't use AI to create patches

Aug 11 • Adrian Sanabria

2

July 2026

I vibe-coded an app for my Pebble watch and it made me sad

I had the opposite reaction to vibe-coding a website, and I think I know why

Jul 13 • Adrian Sanabria

1

1

Signs of hope for Secure by Design

Finally, some potential incentives!

Jul 1 • Adrian Sanabria

2

June 2026

(re)Defining Zero-Day

A vendor patch shouldn’t be part of the definition

Jun 8

2

The unintended consequences of vulnmaxxing

The only way to fix vulns at AI scale is to use AI. Coincidence or cash grab?

Jun 2 • Adrian Sanabria

3

3

May 2026

Verizon's 19th edition of the DBIR confirms the vulnpocalypse***

But with many asterisks! Read on to find out why 😅

May 19 • Adrian Sanabria

3

2