Untrusted Content Masking for Web Agents with Security Guarantees

Defenses that provide security guarantees against prompt injection attacks rely on strict isolation between trusted instructions and untrusted data. In text-based environments such as tool-use APIs, this separation arises naturally: agents can reason from interface definitions without ever...

PT-2026-53521

Content removed...

CVE-2022-30351

PDFZorro PDFZorro Online r20220428 using TCPDF 6.2.5, despite having workflows claiming to correctly remove redacted information from a supplied PDF file, does not properly sanitize this information in all cases, causing redacted information, including images and text embedded in the PDF file, to...