Map / Model Security / Hirundo
Hirundo
hirundo.io ↗ · LinkedIn ↗ · Open on the map
Capabilities
The first machine unlearning platform, removing unwanted data, behaviors, and vulnerabilities from trained AI models without full retraining.
Sensors / integration
Automated evaluations and red-team testing across LLMs and non-generative models (vision, radar, timeseries, speech-to-text, NLP) to surface hallucinations, biases, memorized PII, and jailbreak vulnerabilities.
Protections
Surgically edits model parameters to unlearn poisoned, malicious, or confidential data, biases, and hallucinated behaviors, de-risking models already in production without degrading performance.
Differentiator
Machine unlearning instead of retraining or filtering — claims up to 85% reduction in jailbreak vulnerabilities, 70% reduction in biases, and 100% removal of fine-tuned PII with no impact on other functionality.
Team & funding
- Founders
- Ben Luria, Michael Leybovich, Oded Shmueli
- Total raised
- $8M
- Investors
- Maverick Ventures Israel, SuperSeed, Alpha Intelligence Capital, Tachles VC, AI.FUND, Plug and Play Tech Center
Other Model Security companies
ActiveFence · Advai · Adversa AI · AI Aramid · Alice · AptaSentry · Arize AI · Attestable · Bosch AIShield · Citadel AI · Cranium AI · DataKrypto · See all →
Last updated · Get the “Listed on” badge