Lakera

Lakera is a Web and API-first AI security platform for organizations building and operating generative AI apps, including AI agents, chatbots, and RAG systems in regulated and enterprise environments.

Key capabilities include:

• Runtime protection for GenAI with real-time detection and blocking of prompt injections, jailbreaks, and other prompt-based attacks
• Data protection controls such as data loss prevention, PII redaction, and policies that can be applied by user, app, and action
• Content moderation to filter inappropriate or toxic outputs in production experiences
• Proactive AI red teaming (Lakera Red) to discover vulnerabilities through direct and indirect attack simulations before deployment

Pricing: Freemium. Community Plan is $0/month (up to 10,000 API requests/month, 8,000 token max prompt size, SaaS-only hosting in the EU, community support). Enterprise Plan offers custom pricing with flexible volumes, configurable prompt sizes, SaaS or self-hosted deployment, and enterprise features.

Notable for its multi-model compatibility (including GPT, Claude, Bard, and custom LLMs), sub-50 ms runtime latency, support for 100+ languages, and threat intelligence informed by Gandalf, described as the world’s largest AI red team. The platform reports 0.01% production false positive rate and .