AI Security
Griffin AI vs Gemini On-Device: Developer Tools
Gemini on-device models are fast and cheap. For the developer-tool layer, they're useful. For the engine-plus-LLM layer, on-device is not the right fit.
Mar 20, 20262 min read
Deep dives, practical guides, and incident analyses from engineers who build Safeguard. No fluff, no vendor FUD — just what you need to ship secure software.
Gemini on-device models are fast and cheap. For the developer-tool layer, they're useful. For the engine-plus-LLM layer, on-device is not the right fit.
The difference between grounded reasoning and hallucinated reasoning is not eloquence — it's citation. A look at how Griffin AI anchors every claim.
Platform engineering teams are becoming the new home for security controls. Here's why that is both promising and risky.
SLSA Level 3 requires hardened builds, verifiable provenance, and isolated build environments. Here is the practical path, not the theoretical one.
DNS cache poisoning is a known attack class with a new application: hijacking software update checks to ship malicious binaries that pass every signature check.
Two dozen AI guardrail vendors in 2023. A much smaller set in 2026. The consolidation has pattern — integrated platforms beat standalone guardrails.
An auto-fix that closes a vulnerability and breaks the build is not a fix. Breaking-change awareness separates auto-PRs that ship from auto-PRs that get reverted.
A fact-based review of the best container image scanners in 2026, comparing Trivy, Grype, Snyk, Prisma Cloud, and Safeguard.sh on accuracy and noise.
Trusted Publishing made PyPI safer, but leaked short-lived OIDC tokens in CI logs kicked off a credential-replay campaign that PyPI, GitHub, and Sonatype all tracked in 2025.
Weekly insights on software supply chain security, delivered to your inbox.