Chinese AI models: evaluate cost and security before switching
Lower AI prices create more options. An enterprise decision needs evidence about outcomes, data exposure and the ability to change providers safely.
Blog
Practical guides on penetration testing, vulnerability research, and AI-powered security tools.
Articles
184 articles
LG webOS vulnerabilities: testing Smart TVs on a corporate network
Meeting room televisions need owners and access boundaries too. This analysis turns webOS research into an inventory, update and network testing workflow with observable evidence.
Docker Hub GhostPat: reviewing tokens and trust in container images
A Docker Hub fix highlights the importance of knowing who can publish container images. An original review workflow for credentials, ownership changes and integrity evidence throughout software delivery.
AI replication on IoT: what to test after the Anthropic study
The study investigates capabilities in a simplified environment. Pentesty’s analysis turns the discussion into verifiable questions about connected devices, agent autonomy and access boundaries.
Semiconductor supply chains: the risk a supplier register can hide
A supplier's headquarters does not reveal every equipment dependency. Map support, replacement paths and alternatives before an urgent migration.
AI governance: connect costs, permissions and evidence of use
A spending dashboard cannot explain AI use on its own. Connect identity, purpose, budget and outcomes so deviations can be investigated.
DeepSeek context compression: the reliability test missing from your pilot
Average accuracy can hide retrieval failures. Turn the DeepSeek study into a useful evaluation for enterprise systems and their evidence requirements.
Technical team continuity: who can operate when the specialist is away?
Use the PERM announcement as a prompt to review dependence on individuals, without inferring risk from a worker's nationality.
GhostAction on GitHub Actions: protecting secrets and reviewing workflows
An automation with a security label can still carry dangerous behavior. Our analysis proposes authorized tests for workflow changes, secret exposure and runner boundaries.