Release archive
3.2.126

CRYPTO-PDF-AI-NARRATIVE-SAFETY (Claim Sanitization)

Release v3.2.126 focuses on CRYPTO-PDF-AI-NARRATIVE-SAFETY (Claim Sanitization).

ImprovedInternal
CRYPTO-PDF-AI-NARRATIVE-SAFETY (Claim Sanitization)

Release digest

Improved

  • CRYPTO-PDF-AI-NARRATIVE-SAFETY (Claim Sanitization) — Hardened crypto report AI narrative normalization against standalone buy/sell language, financial-advice framing, guaranteed-return claims, guaranteed-safety copy, and unsupported fully-audited wording.

Internal

  • Extended PDF report contract tests to verify unsafe model output is sanitized both before narrative return and again at the PDF render boundary.
  • Updated docs/architecture/CRYPTO_REPORTS.md because this changes the AI narrative safety contract. Public-docs decision: no user-facing docs change required because report content remains legally safer under the same product behavior.

Technical notes