Today we are excited to roll out PAIG Cloud 1.14, a focused release that strengthens content governance, simplifies reporting, and sharpens enforcement logic across your AI estate. Here is an overview of what we released.
1. Guardrail Framework — Enterprise-Grade Safety & Compliance
Our biggest enhancement adds a native Guardrail framework that lets you codify and enforce responsible-AI policies at scale:
- Content moderation policies – screen every AI response against corporate and regulatory rules.
Denied topics & restricted terms – block disallowed subjects before they reach end users. - Prompt-injection detection – automatically spot and neutralize malicious prompts.
- Automatic PII redaction – strip sensitive data from responses in real time.
- Per-application assignment – apply unique guardrails to each tenant or AI application.
- Custom fallback messages – define user-friendly text when content is blocked.
- Audit logs – review every moderation event for accountability and compliance evidence.
2. Cleaner Report Emails
Scheduled-report emails now use updated templates that remove duplicate text, delivering a crisper, easier-to-scan format for weekly or monthly analytics.
3. New “TOXIC” Tag
You can now label and govern toxic content explicitly. Use the new tag to:
- Build granular policies targeting hateful or harassing language.
- Configure guardrails for automatic detection and handling.
- Standardize toxic-content controls across multiple applications.
4. Policy-Enforcement Fix
We resolved an integration gap between Content Restriction policies and AI application-level rules. Both policy types now evaluate and enforce in concert, ensuring airtight access control.
Availability
As a PAIG Cloud customer, you’re already on 1.14—no action required. Sign in to explore the new Guardrail settings, update your report schedules, and start tagging toxic content today.
Questions or feedback? Reach out to your customer success manager or drop us a note at [email protected].