Startup Diveplane is trying to fix privacy issues around data used in artificial intelligence training with the launch of what it says is the industry’s “first verifiable twin dataset.”

The company’s product, called GEMINAI, is meant to help organizations sell, share and analyze sensitive datasets without worrying about any of the information being lost or stolen or otherwise misused.

The twin dataset it creates can be used for data modeling and analysis just like the originals, but the difference is that it replaces all personally identifiable information with synthetic data that maintains the statistical relationships and nuances contained in the original dataset.

That’s notably different from other privacy techniques that take data and simply mask certain slices of information, such as names and social security numbers, leaving the data vulnerable to conversion back to its original state.

Diveplane reckons GEMINAI will help ensure businesses can comply with privacy laws and compliance regulations such as the General Data Protection Regulation and the Health Insurance Portability and Accountability Act of 1996, known as HIPAA for short.

Diveplane Chief Executive Officer Michael Capps said his company is trying to tackle a big problem in AI. He said many companies find themselves forced to use inaccurate or incomplete datasets when training AI models because they need to satisfy strict privacy requirements. But using incomplete data often means that the AI makes poor decisions, simply because it doesn’t have enough information to come to a correct conclusion.

“With GEMINAI, we’re eliminating that risk by creating a verifiable synthetic ‘twin’ of the dataset, so that businesses don’t need to sacrifice the quality of their AI for the sake of privacy,” Capps said. “GEMINAI offers the best of both worlds and we’re excited to introduce this first-of-its-kind technology to the market.”

Diveplane said it has already seen big interest in its synthetic twin datasets from industries that are most constrained by privacy requirements. For example, hospitals can use GEMINAI to share “truly anonymized patient records” with research organizations to aid the discovery of new medicines and treatments.

GEMINAI can also help to facilitate the multibillion-dollar data-sharing industry, giving companies a way to anonymize information so they can sell it to advertisers, for example, without any repercussions about privacy.

Image: TBIT/Pixabay