CorrectGuard: Eyes-Off Correctness Estimation for Black-Box Security Guardrails

AI services increasingly rely on black-box security guardrails, yet privacy-preserving model auditing regimes often cannot measure how well these systems perform in both a human eyes-off production setting, which disallows human inspection of user input, and a machine eyes-off setting, which...

EUVD-2026-68443

@hulumi/policies versions before 1.3.2 contain an evidence validation bypass vulnerability in Cloudflare and deployment-governance validators that allows attackers to suppress violations by submitting unrelated compliant evidence. Attackers can use evidence from different zones, hostnames, origin...

CVE-2026-82855 @hulumi/policies before 1.3.2 Evidence Validation Bypass

@hulumi/policies versions before 1.3.2 contain an evidence validation bypass vulnerability in Cloudflare and deployment-governance validators that allows attackers to suppress violations by submitting unrelated compliant evidence. Attackers can use evidence from different zones, hostnames, origin...

CVE-2026-82855: Protection Mechanism Failure

@hulumi/policies versions before 1.3.2 contain an evidence validation bypass vulnerability in Cloudflare and deployment-governance validators that allows attackers to suppress violations by submitting unrelated compliant evidence. Attackers can use evidence from different zones, hostnames, origin...

Trust but Verify? Uncovering the Security Debt of Autonomous Coding Agents

The increasing adoption of autonomous coding agents accelerates software development but also introduces scoped security risks within high-impact file paths that can outpace traditional human review capacity. While prior research has primarily evaluated these systems in terms of functional...

poc-factory-sample-output

Prompt Injection Guardrails Introduction In the rapidly e...

Azure Service Tags Vulnerability: Microsoft Warns of Potential Abuse by Hackers

Microsoft is warning about the potential abuse of Azure Service Tags by malicious actors to forge requests from a trusted service and get around firewall rules, thereby allowing them to gain unauthorized access to cloud resources. "This case does highlight an inherent risk in using service tags a...

FTC Fines Mental Health Startup Cerebral $7 Million for Major Privacy Violations

The U.S. Federal Trade Commission FTC has ordered mental telehealth company Cerebral from using or disclosing personal medical data for advertising purposes. It has also been fined more than $7 million over charges that it revealed users' sensitive personal health information and other data to...

U.K. Electoral Commission Breach Exposes Voter Data of 40 Million Britons

The U.K. Electoral Commission on Tuesday disclosed a "complex" cyber attack on its systems that went undetected for over a year, allowing the threat actors to access years worth of voter data belonging to 40 million people. "The incident was identified in October 2022 after suspicious activity wa...