Trending Articles
Trending Topics
Trending
This article is part of our Opinions section, where we invite industry professionals to share their views on the most pressing technology questions of our time.
Before the tsunami of AI put companies on the hunt for broader and deeper training sources, data was recognised as an organisational asset. Even if cross-functional teams weren’t completely ready or equipped to utilise it, there was a broad understanding and acknowledgement that internal data holdings were filled with extractable value. While at first glance the most likely recipient of that value is the organisation itself — advantage delivered in the form of process efficiencies, customer behavioural insights, and repeatable best practices — data’s value is increasingly showing up in ways and places that are unexpected.
In a global, digital economy, insights can come from anywhere. Relevant data often resides both inside and outside an organisation’s walls and across a myriad of silos. It can come from commercial sources, open-source datasets, third-party consortiums, and even competitors. It can also be discovered in cross-boundary, internal assets where the silos and barriers relating to regulatory, legal, or security considerations may limit data utilisation. In order to unlock the value from such sources, analysts and business leaders need to be able to use data across these boundaries while still respecting the reasons they exist.
When leveraging external data sources or datasets located across internal silos, it is essential to consider potential organisational risks by being mindful of data usage barriers as well as the full data processing lifecycle. The most effective way to address these challenges is by ensuring the privacy and security of data while it’s used or processed. This is achieved through the use of Privacy Enhancing Technologies (PETs), a family of technologies that secure the usage of data by enabling, preserving, and enhancing the privacy of data.
While PETs have been the subject of research for decades, recent advancements in efficiency and performance — as well as educational efforts that have broadened exposure beyond academic and research circles — have accelerated the use of these technologies in the commercial and public sector. As a proof point of this advancement, the Information Commissioner’s Office (ICO) in the UK recently led the development of PETs Guidance and other efforts tackling PETs adoption. Elsewhere on the global stage, Singapore’s government launched three initiatives earlier this month, including a PETs Adoption Guide and sandbox, to help advance the use of PETs for AI use cases.
PETs are important because they allow data to be utilised in ways that were not previously possible. They enable users to search and analyse cross-boundary and third-party data sources, expanding the ways in which data can deliver value for a wide range of use cases. To help showcase the impact of PETs, we will highlight three examples of technology-enabled data value extraction made possible by technologies from the category.
The first featured technology from the PETs family is homomorphic encryption (HE). Known for its ability to enable computations in the encrypted or ciphertext space, HE is sometimes referred to as the “holy grail of encryption”, as it allows users to harness the value from data without increasing risk.
One example of how HE can be leveraged to advance business outcomes is its ability to allow users to perform encrypted searches or analytics on datasets across jurisdictional boundaries, relevant in a number of regulated industries, including financial services.
In order to assess potential and existing customers, banks need to be able to utilise data from all branches, including those in other regions or countries. Privacy regulations and data localisation laws frequently limit or block the use of such data, leaving analysts to make decisions based on incomplete information.
By ensuring the content of the search remains encrypted throughout the processing lifecycle, HE allows financial institutions to perform Customer Due Diligence (CDD) and Know Your Customer (KYC) inquiries while respecting existing barriers. Decision makers gain access to a broader, richer collection of data without elevating organisational risk.
Next up is secure multiparty computation (SMPC), a protocol that allows multiple parties to jointly operate on data while their individual inputs remain private. While the security of SMPC can vary widely depending on the type used, it can be implemented in conjunction with other PETs, such as HE, to help amplify the privacy guarantees for a given use case.
SMPC is particularly useful in collaboration scenarios where parties do not want to disclose their private data. For example, healthcare providers looking to improve the quality of their services may want to work together to perform analysis across their respective data holdings. While patient privacy and other regulatory barriers might traditionally hamper such efforts, the use of SMPC allows the providers to identify trends, build models, and test hypotheses across the collective data without exposing individual records. Data can be encrypted and analysed as a whole, delivering only the outputs of interest to the collaborating providers.
The final example of business-enabling capability provided by a technology from the PETs category comes from the use of trusted execution environments (TEEs). Also known as Confidential Computing, TEEs are essentially a perimeter-based security model that exists within a hardware chip.
Although the security posture is weaker than the other examples given, TEEs enable very fast computation, which makes them a solid choice for large-scale use cases, including AI.
Organisations can utilise TEEs to securely train and leverage AI/ML models while ensuring the model development process, the model itself, and the interests of all parties involved remain protected. By enabling users to expand the type of data that can be securely used for training purposes, TEEs help deliver better, smarter models that can enrich outcomes.
The world is constantly producing newer, better, richer data, and organisations need to be ready — and able — to take advantage of assets and sources as they emerge. Users in both the commercial and public sector need capabilities that allow them to leverage data across internal silos and in collaboration with third parties while remaining appropriately mindful of their own interests.
Privacy Enhancing Technologies uniquely protect the data while it’s being used, enabling users to unlock value across a variety of sources without sacrificing privacy or security. By utilising PETs, organisations can securely extract value, collaborate across boundaries, and enrich AI tools at a scale that aligns with our digital age. Harnessing the power of data is essential – and technology is the path to achieving it.
