Research
Forolock research
Understand what AI changes about security.
Research, guides and practical lessons from securing AI-built software and autonomous systems.
Report
The State of AI-Built Application Security
What happens when software creation becomes dramatically faster than security review?
Read reportPublishing soon. Get it by email.
Research and writing.
Findings from our own testing, and practical guides for building safely with AI.
6 articles
Agent Security
Why coding agents should never retain permanent cloud credentials.
Coming soon
Agent Security
Understanding indirect prompt injection.
Coming soon
App Security
How tenant isolation fails in AI-generated SaaS applications.
Coming soon
Guides
Securing MCP servers and tools.
Coming soon
Engineering
Why security scanners create too much noise.
Coming soon
Forolock Lab.
Our research group studies the attack surfaces that AI creates, and publishes what it learns.
AI changes how software fails. The Lab tests real patterns in AI-generated applications and agent systems, reproduces the attacks, and turns what it learns into better defaults for everyone.
Vulnerability research
New classes of weakness in AI-built software and agent tooling.
Status: Planned
Exploit reproductions
Step-by-step reproductions that show how an issue is used.
Status: Planned
Benchmarks
Repeatable measurements of how secure AI-generated code is.
Status: Planned
Test applications
Deliberately vulnerable apps for learning and for testing tools.
Status: Planned
Open-source tools
Small, useful tools released for the community.
Status: Planned