Lakera Guard

service

proprietary SaaS API(no self-host)

note

vendor Lakera acquired by Check Point Nov 2025

source

closed

license

proprietary

Closed SaaS security product; no weights or source. (Lakera also runs the public Gandalf prompt-injection game, but Guard itself is proprietary.) The site is structured around a three-product platform - Workforce AI Security, AI Agent Security and AI Red Teaming - and no longer mentions the Check Point acquisition on the front page; neither moves the score.

• https://www.lakera.ai/ recorded 2026-08-13

  Commercial platform sold through 'Book a demo' and 'Talk to Sales', described under Integration & Scale as 'Easy deployment and scaling', 'API-first architecture', 'Cloud-native deployment'. No source repository, no weights and no self-hostable build is offered anywhere on the site.

An established prompt-injection-defense vendor with notable enterprise customers, acquired by Check Point in 2025. Lakera publishes no customer count, no request volume and no revenue figure; the only numbers on the site are a '1M+ hackers' learning claim from the Gandalf game and 1,000 members of the vendor's Slack community, neither of which counts users of the product. So the level rests on reported standing, and no numeric reach is recorded.

• https://www.lakera.ai/ recorded 2026-08-13

  Claims 'Trusted by the world's leading AI companies' and 'trusted by industry leaders, from Fortune 500 companies to startups' with unattributed pull-quotes, and no customer count anywhere. The only figures are 'Learning from 1M+ hackers around the world' and a Slack community 'With over 1,000 members'.

Category-defining prompt-injection defense; strong coverage, now part of Check Point's platform. The runtime defense now sits beside a red-teaming product and a workforce/shadow-AI product.

• https://www.lakera.ai/ recorded 2026-08-13

  AI Agent Security is described as 'Runtime protection for AI applications' with 'Real-time threat detection', 'Prompt attack prevention' and 'Data leakage protection'; the risks listed are prompt injection attacks, AI data leaks, toxic content generation, indirect prompt injection and multilingual/multimodal attacks, at 'sub-50 ms runtime latency'.