Lakera

Lakera is an AI-native security platform that protects generative AI and LLM applications in real time against prompt injection, jailbreaks, data leakage, and agent-based attacks. Its core products are Lakera Guard (runtime protection) and Lakera Red (AI red teaming). Lakera was acquired by Check Point Software Technologies in 2025.

Founded in 2021 and dual-headquartered in Zurich, Switzerland and San Francisco, Lakera provides an AI-native security platform for enterprises deploying generative AI. Its flagship product, Lakera Guard, is an API-first runtime layer that screens inputs and outputs of LLM applications and AI agents to block prompt injection, jailbreaks, PII/data leakage, and harmful content across 100+ languages with sub-50ms latency. Lakera Red delivers pre-deployment AI red teaming and adversarial vulnerability testing. The company also runs Gandalf, a widely used adversarial-AI education game that has generated tens of millions of attack patterns feeding Lakera's threat intelligence. On September 16, 2025, Check Point Software Technologies announced it would acquire Lakera AI AG in a deal reported at approximately $300 million; the acquisition closed on October 22, 2025. Lakera now forms the foundation of Check Point's Global Center of Excellence for AI Security, and its products are being integrated into Check Point's platform as Check Point AI Agent Security and Check Point AI Red Teaming.

Category

Governance & Security

Pricing

Free tier, Enterprise, Custom pricing

Target Market

Enterprise, Mid-market, Developers

Founded

2021

Headquarters

Zurich, Switzerland (dual HQ with San Francisco, USA)

• ✓
  Lakera Guard

  Real-time, API-first runtime protection that screens LLM inputs and outputs to detect and block prompt injection, jailbreaks, PII/data leakage, and toxic content with sub-50ms latency across 100+ languages.

• ✓
  Lakera Red

  AI red teaming and adversarial testing that runs attack simulations against LLM applications and agents to surface vulnerabilities before production, with risk-based prioritization and remediation guidance.

• ✓
  AI Agent Security

  Runtime protection for autonomous AI agents, including real-time threat detection and prompt-attack prevention for multimodal and multi-step agentic workflows.

• ✓
  Model-agnostic integration

  Works with any LLM (OpenAI, Anthropic, Google, Azure OpenAI, AWS Bedrock, or self-hosted) via a single REST endpoint and SDKs, with central policy control and no application code changes required.

• ✓
  Gandalf threat intelligence

  Adversarial intelligence sourced from the Gandalf education game and a global community of security researchers, producing tens of millions of real-world attack patterns that continuously improve detection.

• •
  Prompt injection and jailbreak defense

  Protect production LLM apps and chatbots from direct and indirect prompt injection, jailbreaks, and instruction-override attacks in real time.

• •
  Data leakage and PII protection

  Detect and prevent leakage of sensitive data and PII in prompts and model responses to meet security and compliance requirements.

• •
  Pre-deployment AI red teaming

  Run adversarial testing against LLM applications and agents to identify safety, security, and responsible-AI vulnerabilities before launch.

• •
  Securing enterprise AI agents

  Add runtime guardrails and monitoring to autonomous and multimodal AI agents operating across enterprise workflows.

Lakera Guard offers a free Community tier (publicly documented as limited monthly API requests) for developers, plus Pro and Enterprise plans. Paid Pro and Enterprise pricing is not publicly disclosed and is quoted via contact-sales; Enterprise adds options such as self-hosting, SSO, and premium support.