21 AI Security & Safety AI tools
AI services in the AI Security & Safety field. Click a name to see its description, pricing and alternatives.
AI Security & Safety: free vs paid at a glance
4 tools with official pricing data. 4 have a free plan, 1 offer a Korean UI.
| Tool | Free | Cheapest monthly | Korean | Where |
|---|---|---|---|---|
| Garak | Yes | — | No | CLI |
| OWASP Top 10 for LLMs | Yes | — | No | Web |
| Shieldstral | Yes | — | No | — |
| TensorFlow Privacy | Yes | — | Yes | API |
From each tool's official pricing page (checked by 2026-10-14). '—' means not stated. Prices change, so check before paying.
How to choose
- Try a tool with a free plan first and check the limit before paying.
- If you need Korean menus, filter by the Korean column.
- Annual plans are often cheaper; the table shows monthly prices.
ART
Adversarial Robustness Toolbox by IBM
Read the guide →Free + paidEU AI Act
EU's comprehensive regulatory framework for AI
Read the guide →Free + paidFoolbox
Fast adversarial attacks for deep neural networks
Read the guide →Free + paidGarak
LLM vulnerability scanner and red-teaming framework
Read the guide →Free + paidHarmBench
Standardized evaluation framework for LLM robustness
Read the guide →FreeLlamaGuard
Meta's LLM-based input/output safeguard
Read the guide →Free + paidMicrosoft Presidio
PII detection and anonymization for text and images
Read the guide →Free + paidNeMo Guardrails
NVIDIA's toolkit for adding safety guardrails to LLMs
Read the guide →Free + paidNIST AI RMF
NIST AI Risk Management Framework
Read the guide →Free + paidOpenDP
Differential privacy library for data analysis
Read the guide →Free + paidOWASP Top 10 for LLMs
Top 10 security risks for LLM applications
Read the guide →Free + paidPySyft
Federated learning and privacy-preserving ML
Read the guide →FreeShieldstral
Mistral's open-weights multimodal safety classifier for text and images
Read the guide →Free + paidTensorFlow Privacy
Training ML models with differential privacy
Read the guide →Free + paidTextAttack
NLP adversarial attacks, augmentation, and training
Read the guide →Free + paidCleverHans
Library for benchmarking adversarial ML attacks
View on the 3D map →Free + paidGuardrails AI
Framework for adding guardrails to LLM outputs
View on the 3D map →Free + paidLakera Guard
Real-time prompt injection and jailbreak protection
View on the 3D map →Free + paidPatronus AI
Enterprise LLM evaluation and safety platform
View on the 3D map →Free + paidPrompt Shield
Microsoft Azure's prompt injection detection service
View on the 3D map →Free + paidPyRIT
Python Risk Identification Toolkit for GenAI by Microsoft
View on the 3D map →