About Pacific AI Labs
Security, privacy & governance for the age of AI-written code.
We build the layer that checks what AI produces — before it runs. We started because the vulnerabilities AI ships are the oldest, most preventable ones, and the reviewer who used to catch them is gone.
How we build
Using AI to build a security tool only helps if the way they work together is disciplined. Ours runs on a few hard rules:
- Three models, checking each other. Different vendors (Claude, Gemini, GPT) build and red-team the work — no single AI is reliable enough to trust alone.
- Genuinely isolated. Each model runs in a separate, least-privilege environment, so steering one can't reach the others.
- Live re-execution, not pasted output. A claim counts only when the check is re-run and seen to pass.
- A human decides. Findings are staged as recommendations; nothing ships without a person's explicit sign-off.
That's governance, not just security: who may decide, what's recorded, what's auditable — and a single accountable human holding the final yes. Independent review by these models found four bypass classes our own testing missed; all are fixed and kept as permanent tests.
The founder
Amanda Malave
Founder & Safety Officer
Security and governance leader with 18+ years across Google, T-Mobile and the University of Hawaiʻi. Most recently the lead security and governance consultant to Google Labs and Google Research, where she owned her team's vulnerability management. Former principal security program leader at T-Mobile, and a lecturer in ethics, digital marketing, and global strategy at the University of Hawaiʻi.
Based in Kapolei, Oʻahu, Hawaiʻi — a developer, creator, mother, and dog lover.
Her philosophy runs through everything here: AI can build fast, but a human decides what ships — and nothing ships unchecked.
Let’s build with AI, safely.
Explore the Security MCP use case, or start a conversation about consulting and custom builds.