From left: Jack Fitzsimons and Robert Pisarczyk. Image: Oblivious
Founded by two Oxford PhD graduates, Oblivious is developing novel technologies with the aim to make data privacy the default and ‘not just an afterthought’.
Data is the building block of the modern digital economy. But one of the biggest challenges around data is walking the tightrope between using valuable information to gain business insights and respecting the privacy rights of individuals.
Laws around the fair and just use of data that prioritises privacy have taken shape substantially over the past decade, especially with the introduction of the GDPR in Europe. Companies around the world are required to follow a strict set of rules and regulations when handling sensitive data.
For many, however, walking this tightrope can be challenging.
“We have found that the world’s impactful data exists in silos across one or more organisations and data scientists are often unable to do their jobs because of the restricted data access,” says Jack Fitzsimons, co-founder and chief technology officer of confidential computing start-up Oblivious.
Privacy enhancing technologies
Based in Dublin, Oblivious builds tools that allow data scientists and machine learning models to work on sensitive data while aiming to enforce confidentiality requirements and brokering trust between businesses.
“We want to enable data scientists to work with sensitive data by integrating privacy-enhancing technologies to their existing workflow and help organisations leverage the power of data without compromising privacy,” Fitzsimons adds. “In the era of rapid advancements in machine learning and the growing importance of data, this issue is more pressing than ever.”
Fitzsimons founded Oblivious in 2020 with fellow Oxford University alumnus Robert Pisarczyk. An expert in data science and machine learning, Fitzsimons holds a DPhil (Doctor of Philosophy) from Oxford, where he first began his professional association with Pisarczyk while the two were working on their doctorates.
“I was working on machine learning while Robert was focused on quantum computing and cryptography. We came up with ideas on how to solve the challenges around protecting information while making non-sensitive macro-insights accessible,” Fitzsimons says. “Instead of writing up some more academic papers, we decided to build something ourselves.”
Oblivious works with two core underlying technologies: trusted execution environments, or TEEs, and differential privacy.
The former refers to secure sections in a computer system that keep sensitive data safe and protect information “while it’s being used” so that no unauthorised access is granted. It also tells data providers how the data will be used through its life cycle.
Differential privacy, on the other hand, is a technique that helps maintain the privacy of data subjects “whenever insights need to be extracted”, such as statistics or machine learning models. This ensures that the information shared or analysed does not lead back to any individual personally, while still allowing companies to understand overall trends or patterns.
“When we combine these two techniques, we get a very secure and private way of handling, combining and processing sensitive data,” Fitzsimons explains.
“The information is first sent to and processed safely in a TEE, where it is kept secure from outside interference. Once the data has been processed, differential privacy is applied to the results, protecting it from reverse engineering attacks.
“This combination allows for valuable insights to be derived from data without compromising the privacy of individuals or entities like businesses. This is particularly important for sectors like healthcare or finance, where there’s a lot of sensitive data, but individual privacy needs to be respected.”
Beyond the Oblivious
The technology seems to be paying off for Oblivious. In April, the start-up announced a raise of €5.35m in an oversubscribed seed funding round led by Berlin-based Cavalry Ventures. It has previously raised $1m in a seed round in 2021.
“Collectively, the tech industry has put hundreds of billions into building robust data lakes and advanced data science tooling, but standard role-based access control still remains untouched since the ’80s,” co-founder and CEO Pisarczyk said at the time.
“Put yourself in the shoes of Uber, Airbnb or any modern tech-enabled company. You can’t simply hand your data scientist the keys to customer data. Like with any powerful resource, it could be used for great good or great evil.”
Oblivious has partnerships with the likes of CeADAR, the United Nations, OpenDP, IBM and Microsoft. But Pisarczyk and Fitzsimons don’t just want to stop at simply building the technology.
The start-up has been steadily expanding its team since the beginning of the year and launched Antigranular, a community-driven, open-source platform that allows members to explore privacy-enhancing technologies, collaborate with and learn from other data scientists.
This culminated in the Eyes-Off Data summit earlier this month that saw speakers from across the world come together to discuss the rapid rise of AI and how to be responsible with data using PETs.
“PETs allow us to unlock the power of the world’s most sensitive data without compromising privacy,” Fitzsimons told the conference. “We’re trying to build a world where data respects its boundaries, trust is brokered via reliable technologies and privacy is the default, not just an afterthought.”
10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.
