)
Pankaj Thapa and Dr Aditya Narayana K. Image: Mirror Security
According to the start-up, the funding will be used to expand its engineering and AI security teams in Ireland, the US and India, as well as drive expansion into US enterprise markets.
Dublin-based cyber start-up Mirror Security has announced today (2 November) a pre-seed fundraise of $2.5m to scale its encryption platform for AI security.
The start-up, which was founded by Pankaj Thapa and Dr Aditya Narayana K last year, said it’s addressing one of the “most urgent problems in enterprise AI adoption” – the lack of reliable data confidentiality in model training and inference.
Mirror Security has built a ‘Security of AI’ platform, which includes three main security technologies developed by the start-up: AgentIQ, Discover and VectaX.
AgentIQ provides “full-spectrum agentic security” as well as guardrails and compliance for AI agents, while Discover provides risk assessment and automated red teaming for AI systems.
VectaX, according to Mirror, is the world’s first production-ready fully homomorphic encryption (FHE) engine optimised for AI workloads.
This encryption tech enables AI systems to process sensitive data while keeping it completely encrypted.
“AI is transforming productivity and reshaping industries, but unsecured data flows pose an urgent risk that threatens to stall adoption,” said Thapa, who is also Mirror’s CEO. “With VectaX, we replace policy-based trust with cryptographic proof, ensuring that sensitive information remains encrypted even during computation.
“Our mission is bigger than protecting data – we are building the trust layer for the AI economy.”
Mirror Security, which spun out from University College Dublin, originated from Narayana’s thesis and is backed by 23 team-held patents in cryptography, security and AI security.
The funding round was led by Sure Valley Ventures and Atlantic Bridge, along with support from strategic angel investors, and will be used to expand engineering and AI security teams in Ireland, the US and India, accelerate product development in encrypted inferencing and secure fine-tuning, and drive expansion into US enterprise markets.
“Data security remains one of the most significant hurdles to AI adoption, with organisations concerned about exposing proprietary information during model training and inference,” said Weili Wang, investor at Atlantic Bridge. “Mirror Security’s experienced team and differentiated FHE technology provide a unique solution for secure AI deployment without compromise.
“We are delighted to partner with the team and support this mission through our global network of cybersecurity and AI leaders.”
As well as today’s funding news, Mirror Security has announced a multimillion-dollar strategic agreement with agentic AI solution provider Inception AI.
Under the partnership, Mirror will deploy its full AI security stack across Inception’s enterprise and government ecosystem.
Mirror also previously forged strategic partnerships with Intel, MongoDB, Qdrant, SiSys AI and Accops.
The topic of AI security and data protection has become a major point of discussion recently. In October, a survey of 1,000 Irish office workers by Accenture found that 19pc admitted to inputting sensitive business data such as customer details and financial information into free, unsecured AI tools.