Sunday, October 4, 2026 Sign inSubscribe to The Brief →
Humanoid Threats

Security for the age of physical AI

Directory

Lakera (Check Point)

AI security company focused on protecting LLM applications from attacks such as prompt injection, acquired by Check Point in 2025.

Category
AI and LLM security
Website
www.lakera.ai
Last verified
October 4, 2026

What is Lakera (Check Point)?

Lakera is an AI security company best known for protecting large language model (LLM) applications from attacks such as prompt injection. Check Point acquired Lakera in 2025 to build out its AI security stack.

Key capabilities

  • Runtime protection for LLM-powered applications.
  • Defenses against prompt injection and related attacks on AI assistants.

Why it matters for humanoid robot security

Many humanoids now accept spoken or written instructions processed by language models. Prompt injection, where hidden instructions hijack the model, is one of the most direct ways to misuse an LLM-controlled robot.

Where it fits in a humanoid robot security program

Lakera focuses on the language interface. Many humanoids now take spoken or written instructions that pass through a large language model, and that is where prompt injection and jailbreaks happen. Lakera-style input and output filtering sits between the user and the model that plans the robot's actions.

Who it is for

  • Robotics teams using LLMs or AI agents to plan robot actions
  • Security teams responsible for AI models in production
  • Organizations that already standardize on enterprise AI security tooling

Questions to ask Lakera (Check Point)

  • Can you protect models that run on the robot itself, not only in the cloud?
  • How do you detect prompt injection that arrives through speech, images or sensor data?
  • What latency do your controls add to a robot's decision loop?
  • How do you log and report AI incidents for later investigation?

Similar listings

Sources

This listing is editorial and free; inclusion is not an endorsement. Work at Lakera (Check Point)? Claim or update this listing.

Get the weekly Humanoid Threats Brief → Subscribe

The Humanoid Threats Brief

The weekly briefing on humanoid robot security.

New vulnerabilities, incidents, standards and defenses, with why each one matters. Built for security teams, robotics engineers and the people buying humanoids.

  • Every claim sourced. We link the CVE, the paper or the regulator, not rumors.
  • 5-minute read. One email a week, every Thursday.
  • Free. No spam, unsubscribe in one click.

Check your inbox to confirm your subscription.

Something went wrong. Please try again.

Read a recent storyUnitree G1 Vulnerabilities Explained: CVE-2026-76639 and CVE-2026-76640 →
100% primary-source linked 1 email a week