Hirundo AI’s $8M Investment to Transform Data Privacy with ‘Machine Unlearning’

Hirundo AI, a top startup in its field of “machine unlearning,” has recently closed an $8 million seed round. Maverick Ventures Israel led the investment round, with SuperSeed, Alpha Intelligence Capital, Tachles VC, AI. Plug and Play Tech Center, will empower the launch and commercialization of Hirundo’s transformative technology in the area of data privacy and be used to solve major problems like AI hallucinations and biases.

Machine unlearning is a novel idea that enables AI models to “unlearn” individual data points or behaviors they have previously learned, but without the excessive time and cost of having to fully retrain the overall model. This is becoming especially important in a time when data privacy laws like GDPR’s “right to be forgotten” are in full force and with a growing responsibility to validate the accuracy and reliability of AI systems.

Hirundo AIs innovative engine mitigates the issue of AI hallucinations and biases at the source, rather than covering them up. The company ’s offerings allow enterprises to extract poisoned, malicious, or sensitive data from trained AI models, cutting biases by up to 70% and hallucinations by up to 55% in early pilots. With this “AI model neurosurgery” technique, we can surgically pin-point and delete undesirable behavior from the model’s internal construction – hardening AI systems against adversarial attacks and making them compatible with the law.

The capital will allow Hirundo to expedite product development and add new customer deployments in high-stakes industries such as finance, healthcare, and defense where data precision and regulatory trust are paramount. Hirundo is already testing its product with MNCs, and government organisations for deployment in mission critical AI.

Hirundo AI, established in 2023, is the world’s only machine unlearning platform and the first of its kind startup. This investment is a major step towards a world where AI models are not only more efficient, but also safer and more reliable in deployment, all while remaining compliant with ever-growing data protection requirements.