Editorial Policy

Scope and standards

PermsAI covers AI security, web security, disclosed vulnerabilities, model evaluation, and consequential AI developments. We prioritize material that helps readers understand technical risk and make better engineering decisions.

Sources and fact checking

We prefer standards, official advisories, incident reports, research papers, and original benchmark documentation. Secondary reporting may add context. Material factual claims should be checked against the strongest available source, while uncertainty and inference should be labeled.

Independence and commercial separation

Editorial conclusions are not sold. Advertising or sponsorship, if introduced, will be separated from editorial decision-making and clearly identified.

Benchmarks and vulnerabilities

Benchmark coverage should describe the evaluated task, conditions, version, and limitations rather than treating one score as universal capability. Vulnerability coverage avoids unnecessary exploit detail, links to authoritative advisories, and respects responsible-disclosure constraints.

Corrections

We correct confirmed errors and explain material changes under our Corrections Policy.