Catalog: 10 official guardrails

IBM has published the Granite Guardrails Catalog, a set of 10 open-source, production-ready safety classifiers for detecting and mitigating risks in generative AI outputs, including hate speech, harassment, self-harm intent, misinformation, privacy leaks and PII detection. Each guardrail combines fine-tuned small language models with lexical and contextual checks, runs offline in Python and integrates with Hugging Face Transformers, LangChain and LlamaIndex, with IBM issuing quarterly validation reports covering precision, recall and cross-lingual performance. IBM states the guardrails are technical safeguards rather than legal compliance tools, and they are released under Apache 2.0 with semantic versioning and bias-audit checks in the CI/CD pipeline.

The IBM Granite Guardrails Catalog comprises 10 officially published, production-ready AI safety guardrails—developed and maintained by IBM—for detecting and mitigating risks in generative AI outputs. These guardrails are open, auditable, and designed for integration into enterprise AI applications via IBM watsonx. ibm-granite/guardrails . The IBM Granite Guardrails Catalog is a curated set of 10 production-grade, open-source safety classifiers. Unlike heuristic filters or proprietary black-box models, each guardrail applies transparent, rule-augmented ML logic—combining fine-tuned small language models with lexical, syntactic, and contextual checks. They are not regulatory requirements but engineering controls aligned with NIST AI RMF’s “Govern” and “Map” functions. The 10 guardrails are: hate speech , harassment , sexual content , self harm intent , violence threat , misinformation claim , medical misinformation , financial misinformation , privacy leak , and pii detection . Each is versioned, tested, and documented independently. IBM publishes quarterly validation reports—including precision, recall, and cross-lingual performance metrics—for all 10 guardrails. Testing uses stratified, human-reviewed datasets drawn from real-world user prompts anonymized and consented and adversarial red-teaming corpora. Updates follow semantic versioning e.g., hate speech-v2.3.1 and require CI/CD pipeline approval, including bias audit checks against protected attributes. No guardrail is updated without backward-compatible API contracts and changelog disclosure. They run natively in Python ≥3.9 , integrate with Hugging Face Transformers, LangChain, and LlamaIndex, and deploy via Docker, Kubernetes, or serverless functions. IBM provides prebuilt containers and Terraform modules for AWS, Azure, and IBM Cloud. Importantly, all 10 guardrails operate offline—no telemetry or external API calls required—meeting strict air-gapped and sovereign-cloud compliance needs. A: Não. Eles são opt-in safety controls—enabled per application or prompt flow—and fully configurable in watsonx.governance dashboards. Q: Há suporte a português brasileiro? A: Sim. As versões 2.2+ incluem native Portuguese pt-BR support for all 10 guardrails, validated on BR-specific slang, idioms, and cultural context. Q: Esses guardrails substituem auditoria humana ou conformidade regulatória? A: Não. IBM explicitly states they are technical safeguards , not legal compliance tools—complementing, not replacing, human review and jurisdiction-specific governance. Q: Posso modificar ou estender um guardrail? A: Sim. Under Apache 2.0, users may fork, adapt, and redistribute—provided attribution and license notices are preserved. IBM documents extension patterns in its Guardrails Developer Guide. privacy leak and pii detection guardrails comply with ISO/IEC 27001 Annex A.8.2.3 and align with BCB Resolution 145/2023 on personal data handling in financial AI. misinformation claim Q3 2024 report . a7c3f9d 2024-09-11 Saiba mais em https://g.cloud https://g.cloud Originally published at g.cloud https://g.cloud/blog/en/catalogo-10-guardrails/ — the guardrail every AI answer passes through before reaching a human.