From the lab
Research
Security research and practical engineering guides. Read about AI attacks, application security, penetration testing, and the work of bringing AI into production.
Explore the benchmarksThe Counterweight newsletter
Stay updated.
Get new research, engineering guides, and benchmark releases in your inbox.
Service guides
Work with usAI agents are hunting leaked secrets: what SaaS teams should fix first
A recent Microsoft report connects exposed repository credentials to cloud access. Review the identity behind each secret, its permissions, and your revocation process.
Read the guideCoding-agent supply-chain security: model backdoors and tool permissions
Research on poisoned training data and model checkpoints raises a practical question: what can a coding agent change when its model makes an unsafe decision?
Read the guideThe cost of shift-left security: tooling, triage, and developer time
Budget for the work a security check creates, including triage, remediation, and upkeep. Use an explicit cost model before claiming that moving a check earlier saves money.
Read the guideSecurity requirements for spec-driven development
Turn security requirements into enforceable behavior and negative tests before an AI coding agent starts implementation. A tenant-isolation example shows the full chain.
Read the guideHow to reduce recurring vulnerabilities in a SaaS codebase
Use repeated findings to identify shared design problems. Move from isolated patches to maintained controls, regression coverage, and a measure of recurrence.
Read the guideAI-generated security fixes: how to verify the vulnerability is gone
Review the security rule, test the original failure, and check adjacent paths. A plausible patch and a green test suite are only part of the evidence.
Read the guide
Showing 1–6 of 14
From the lab
PSIRT under active exploitation
The model has to turn an incomplete vulnerability report into verified scope, containment, safe fixes, and accurate customer guidance.
Recovering from ransomware without reviving it
Backups include clean, stale, encrypted, and persistence-tainted states, and services must return in dependency order.
Both sides of the scale
Why we grade a model's offense and defense in the same environments, and what every future result must report.
Showing 7–9 of 9