The UnboundCompute blog

Writing on web security

Juice Jacking: Can a Public USB Port Really Steal Your Data?

Can a public USB port steal your data? Here is how juice jacking works, whether the risk is real today, and the…
SIM Swapping: How Attackers Hijack Your Phone Number

Sim swapping lets attackers steal your phone number and hijack your accounts. Learn the warning signs and the steps that stop it…
The Fine Tuning Jailbreak: How Training Strips Safety Alignment

A fine tuning jailbreak strips a model’s safety with a few training examples. How the attack works, why alignment is fragile, and…
Insecure Output Handling: When Apps Trust the Model’s Words

Insecure output handling is the OWASP LLM bug where apps trust model text into a browser, shell, or database. See how it…
Excessive Agency in AI Agents: The Risk That Turns a Trick Into a Breach

Excessive agency is why a small prompt injection becomes a real breach. Learn its three parts, how to spot it, and how…
Model Extraction Attack: Stealing a Model Through Its API

A model extraction attack copies a paid model through its API alone. Learn how query based stealing works, how to spot it,…
Membership Inference Attack: Proving a Record Was in Training Data

A membership inference attack proves one record was in a model’s training data. Learn the memorization signal it exploits, and how to…
The AI Prompt Injection Worm That Spreads Between Agents

An ai prompt injection worm hijacks one AI agent, then copies itself into the output so the next agent gets infected. See…
The Policy Puppetry Attack: When User Text Pretends to Be System Policy

The policy puppetry attack disguises user input as official config so a model obeys it as policy. Learn why it works, how…
Skeleton Key: The Jailbreak That Rewrites a Model’s Own Rules

The skeleton key jailbreak does not break a model’s rules, it persuades the model to rewrite them so it answers with a…