Essays
Compliance writing usually reads like a brochure. Ours doesn't.
Each piece starts with a scenario a real deployer faces, argues from how the technology actually behaves, and shows the real output — including a failing case — rather than a marketing screenshot. Most importantly, every essay states plainly what CloakLLM does not do. Publishing the limits is what makes a compliance claim worth trusting; it's the throughline of everything below.
Hand your logs to an auditor — they can check them without us
The standalone verifier that makes “verify, don’t trust” literal: an auditor re-checks the hash chain, timestamps, key provenance, and compliance reports on their own machine — without our SDK, and without trusting our code. Includes what a passing result does NOT prove.
ReadCompliance you can verify, not just believe
Why GDPR and the EU AI Act can’t both be satisfied by a normal logging pipeline — the Article 12 Paradox — and the honest, measurable way CloakLLM resolves it. With the real detection numbers, including the ones that aren’t flattering.
ReadThe Article 12 Paradox
The foundational whitepaper: log everything (EU AI Act) versus keep nothing (GDPR), and why stripping PII before it reaches any log is the only way to satisfy both obligations at once.
ReadUsing CloakLLM? Preparing for a 2027 audit?
We're talking to teams deploying AI under the EU AI Act — what you're building, what an audit will ask of you, and what's missing. Auditors and certification bodies especially welcome.
Talk to us