Founded in 2022, this Austrian start-up has built infrastructure that enables companies to compute on sensitive data without ever decrypting it.
Data security is one of the most important considerations in the world of tech. With breaches and cyberattacks more prevalent than ever, protecting sensitive data has become a top priority in every industry.
As efforts to maintain and improve cryptography standards continue to progress, a team based in Austria’s Graz University of Technology realised there was a problem that came with advanced encryption – in order to collaborate using this sensitive data, companies would need to decrypt it first.
“For over a decade, industries have needed to compute across private, distributed datasets – yet privacy and collaboration have never co-existed,” says Lukas Helminger, a former cryptography researcher and lecturer at the university. “We believe this is the internet’s biggest unsolved problem and it is our goal to solve it.”
Helminger, along with a team of fellow cryptographers – Daniel Kales, Roman Walch, Lukas Götz, Stefan Plank and Christian Rechberger – decided to establish our Start-up of the Week in 2022 to do just that.
Taceo is an Austrian start-up that builds cryptographic infrastructure to enable companies to verify and compute on sensitive data without ever decrypting it.
“Our technology makes it possible for multiple parties to collaborate securely on private data,” says Helminger, who is also CEO of the company.
How it works
Taceo’s founding team called upon a range of experience in the discipline – having developed cryptographic projects such as the Poseidon hash function and a Covid-19 contact-tracing prototype – to create Taceo’s core product, the Private Shared State.
As Helminger explains to SiliconRepublic.com, the Private Shared State is a data model that lets multiple parties jointly compute and verify information across encrypted data without ever revealing the raw inputs.
The Private Shared State is powered by a cryptographic tool called coSNARKs, which combines two advanced cryptographic techniques: multiparty computation (MPC) and zero-knowledge proofs (ZKP).
MPC allows multiple participants to compute together on encrypted data ‘shares’ without revealing their individual inputs, while ZKP allows a participant to prove something is true – such as a transaction – without sharing the underlying data.
“Both these methods combined allow multiple parties to jointly prove a fact/statement is true, using minimal data and no back-and-forth communication,” says Helminger. “Together, these approaches enable fast, collaborative and verifiable computation – with privacy and security baked in by default.”
According to Helminger, Taceo’s immediate focus is on the financial services sector, where privacy-preserving infrastructure is urgently needed for compliance, payments and identity verification.
“Longer term, we’re building towards enabling encrypted compute for other applications, such as AI, which would allow LLMs to train on or verify sensitive data without risking privacy or IP,” he says.
How it’s going
Since hitting the start-up scene in 2022, Taceo has achieved a variety of milestones. The company previously partnered with cryptocurrency project World (formerly Worldcoin) to deploy a system for privacy-preserving biometric verification, protecting the iris data of more than 14m people according to Helminger.
And, at the end of July the start-up raised €4.8m in a seed funding round led by Archetype VC with participation from a16z CSX, Cyber.Fund, A.Capital Ventures, Polymorphic Capital and a number of prominent angel investors.
“This funding will help us grow our team from 12 to [approximately] 20 and accelerate the development of encrypted compute tools for AI,” says Helminger.
Not unlike most tech ventures, Taceo has experienced its fair share of challenges. Unique to the start-up, however, are challenges that revolve around privacy and security.
“Privacy is a loaded term and is often not understood or positioned correctly. Most people confuse it with security, secrecy or just a lack of data collection,” says Helminger. “In reality, privacy is a much broader, more nuanced concept about control and autonomy over one’s personal data, even by those who are authorised to have it. This sometimes makes it a challenge to communicate our value prop, which is more about empowering users with control, rather than just hiding information.”
He explains that the stakes are “exceptionally high” for deploying Taceo’s tech, which in turn demands a significantly high standard of security and reliability.
“Unlike AI models that can be shipped at 60pc or 70pc accuracy and incrementally improved, a security protocol must be 100pc secure from the outset,” he says. “You cannot release a ‘mostly secure’ product when handling sensitive, confidential data. A single vulnerability could have catastrophic consequences, which is why there’s no room for error.”
Going forward, Helminger says the company is eyeing further fundraising to fuel its growth.
“Following our recent seed round, we will be planning further investment to expand our infrastructure and support new use cases,” he says. “This is highly scalable technology with broad applications across finance, AI and national security – sectors where the need for secure, trustless infrastructure is only growing.”
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.
