Hirundo
Machine unlearning platform for removing risky behavior from AI models
About Hirundo
Hirundo provides a machine unlearning platform that removes unwanted knowledge and behaviors, such as bias, jailbreak vulnerabilities and memorized personal data, from trained models without retraining. It also offers data quality checks for non-generative AI datasets.
Common use cases
- Removing PII from fine-tuned models
- Jailbreak mitigation
- Bias reduction
Products and services
Behavior Unlearning
Targets jailbreaks, hallucinations, bias and toxicity.
Data Unlearning
Removes unwanted or personal data from models.
Data QA
Quality checks for vision, speech and NLP datasets.
Key features
- Machine unlearning
- Red-team evaluation
- PII removal
- Bias mitigation
- Dataset quality checks
Hirundo alternatives
- LALakera
- PAProtect AI
- RIRobust Intelligence
GiskardTesting and red-teaming platform for AI agents and LLM applications
FAQs
What does it do?
It removes unwanted behaviors and data from trained models without retraining.
Who is it for?
AI, security and privacy teams.
How is it priced?
Not published.
Work at Hirundo? Request updates to this profile. Update this profile
Compare with similar providers
All ML Operations providersActualyze AIAI governance gateway between applications and model providers
AlationData catalog and governance platform for trusted enterprise data and AIALERT AIAI governance, security and cost control inside the customer's VPC
AlignAIAI program management and governance platform for intake and risk review
ArthurControl plane for governing, monitoring and securing enterprise AI agents
AscertaEnterprise AI value measurement and spend tracking platform
Barndoor.aiEnterprise AI gateway governing models, MCP servers and workflows
BraintrustEvaluation and observability platform for AI agents