Patent published on August 10, 2023

SAP's New Way to Guess Passwords: Using the Internet to Understand us Better

In a progressive era powered by bits and bytes, German software company SAP potentially takes cyber security to a new level. With a newly minted patent (US20230252114A1), they are attempting to read our minds on a digital scale. The subject we're addressing is passwords, usually a bane of our Internet existence.

In an attempt to make passwords more secure, SAP has ventured into the realm of machine learning. This software program, described in patent US20230252114A1, is an optimistic attempt to guess passwords. You may be wondering how it accomplishes this. It takes a deep dive into cyberspace, mining common words, phrases, and even the latest lingo to anticipate what people are inclined to use for their digital locks.

Previously, trying to foresee what character combinations people would employ for their passwords was a bit like aiming darts in the dark. Researchers often had to assemble extensive lists of potential password vocabulary. Not only was this method time-consuming but it also required intimate familiarity with specific themes, such as the terminologies of a new game. This new solution, according to SAP’s patent, would use the wealth of information already available online, saving researchers valuable time and effort.

Traditional password strength indicators often rely on checking rule-based criteria or entropy measures. However, SAP argues that with the fast-paced evolution of digital culture and its impact on password creation, a more automatic, yet smart, system is required.

We should note the advantages of such a system. Rather than focusing on a basic mix of letters, symbols and digits, this more advanced approach incorporates semantic categories. These categories, derived from the accessible online language, are then utilized in unsupervised learning methods. It means the model can be more flexible and well-adjusted, whilst maintaining the computational resources at a low level.

The diagrams included with the patent (figures 1 through 6) illustrate in broad terms how machine learning would be used to generate password guesses. The structural configurations and comprehensive methodologies discussed affirm the potential of the patent's proposed approach.

But, as with every shiny new invention on paper, it's yet unconfirmed whether this program will actually make it to the market. For now, it remains a promise in a patent, hinting towards a future where our digital security would be less of a guessing game and more of a sophisticated match of wits against an intelligently learning machine. This application might soon appear in SAP's leading product, SAP HANA, but it's too early to make any predictions. Till then, we keep our fingers on the keyboards, creating the best passwords we can muster.

Explore more