With the new PET-based solution, it is possible to collaborate on data across departments on a real time basis to deal with crime, tax evasion, and stock market manipulation.
| Photo Credit: Getty Images
Any transaction, online or offline, involves sharing of personal data like name, date of birth and other details. Despite fears of privacy breach, there is little anyone can do, but hope the information is not misused, since it has become mandatory to disclose such information many times.
Unique software algorithm
While the Data Protection Bill is still in the making, a city-based software firm has unveiled a unique software algorithm where personal data can be ‘shared’ between organisations without breaching the privacy wall.
Posidex Technologies, a deeptech AI/ML product company, has come out with a Privacy Enhancing Technology (PET) promising a paradigm shift on how different entities like financial institutions or government departments can ‘collaborate’ on data meeting, data privacy and localisation norms.
The firm co-founded by classmates – Bhavani Shanker Chittor, K. Venkat Reddy, G. Venugopal Rao, alumni of CBIT, KITS, NIT Warangal, respectively, in 2003, has launched a new algorithm that converts plain text data into “Vectorised, Anonymised, Randomised, Encrypted Tokens” or VARE Tokens, on which an AI/ML model is built adding ‘intelligence’ that allows data comparison.
There are solutions like Zero Knowledge Proof (ZKP), Homomorphic Encryption (HE), Secure Multi Party Computation (SMPC), and Differential Privacy, all part of emerging cryptography based PETs, allowing analytics and insights on transformed personal data tokens, said Managing Director G.T. Venkateshwar Rao.
Encrypted token matching
In traditional ‘encrypted token matching’, used in storing of passwords, even if a single letter – capital/small or digit is wrongly entered, the request is rejected as there is no ‘intelligence’ in the tokens to compare. With Posidex tech, a new method of adding ‘intelligence’ to tokens for all data elements, including name and birth date, can be done, he said.
Here customer data is converted to intelligent ‘VARE’ tokens, amenable for probabilistic search with fuzzy logic on all customer data elements, said Mr. Rao, a former IRS officer (1990 batch), recipient of PM’s award for excellence in public administration. He had also worked as MeeSeva Commissioner with Telangana Government (2015 to 2022).
The technology will also be useful for regulatory authorities to collaborate with each other’s data, without actually sharing it. Income Tax, GST, SEBI (Securities & Exchange Board of India) and others, presently share information on a case-to-case basis and are not open to give complete access to their data to each other, he said.
With the new PET-based solution, it is possible to collaborate on data across departments on a real time basis to deal with crime, tax evasion, and stock market manipulation. Data privacy is ensured since the search happens on encrypted data even as it stays with the respective department, said Mr. Rao.
PETs are evolving tech still under testing stage in the US, Israel, France, and other countries, but this firm is sure about its solution being ready for various needs, since the AI model is trained on more than five billion actual customer data over the last 18 years.
The firm has been providing customer insight IT solutions to more than 60 enterprises, including seven of the nine largest private sector banks and departments like I-T, GST, Governments of TG, Odisha and Tamil Nadu, he added.
