Since its release by OpenAI at the end of last November, ChatGPT has been a constant topic of discussion, highlighting the often amazing results or ridiculous mistakes of artificial intelligence. However, the significance of ChatGPT is not given by its current capabilities, but by the huge resources directed towards its rapidly accelerating development, such as Microsoft’s ten billion dollar support for OpenAI in January.
Thanks to these, in the coming years, chatbots powered by artificial intelligence can undergo a development similar to that predicted by Moore’s Law for integrated circuits since the 1960s.
Tech companies pouring billions of dollars into artificial intelligence research ultimately want to generate profit, opening the door for software like ChatGPT to appear in a variety of products and services from other companies, from car information systems to interactive travel guides on smartphones.
One such use for ChatGPT could be to strengthen password protection for average users, which can never be secure enough despite commonly used password managers like LastPass and the increasingly required two-step authentication.
.png) |
LastPass Password Manager
(Credit: LastPass) |
Since today the average person leaves a significant digital footprint of their life, from family photos to profiles created on social networks to utility bills paid over the internet, a ChatGPT module specifically optimized for this and integrated into password manager software could create a unique database of verification questions for each user that is truly personalized and goes beyond basic questions like the name of the person’s first pet.
By running this ChatGPT module on its own personal material, artificial intelligence can compile a list of up to a hundred facts for the given user, which the user can then individually verify, confirming their authenticity and usability in the password manager software.
These personalized verification questions can be based not only on general data such as place of birth or the name of the person’s kindergarten, but also on completely unique information derived from processing family photos, such as the nickname of a favorite uncle or the location of a camping trip at the end of summer 2010.
Although it is obviously possible to compile such a list of facts without the help of artificial intelligence, with machine assistance it can be done in minutes instead of hours, as the user does not have to search for data and browse through thousands of old photos, just check the ready-made list of facts. ChatGPT can also associate an image with a purely textual data such as graduation year by searching for an image of the person’s school on the internet if no suitable image is found in the user’s own collection.
.png) |
reCAPTCHA test (Credit: Alphabet) |
As a result, a new level of verification can be created and integrated into Norton Password Manager and other similar password manager software based on these fact lists, which at first glance resembles image CAPTCHA tests that precede automatic password entry, which is truly personalized and does not require a mobile phone unlike two-step authentication using SMS.
With these tests based on personal data and memories, password manager software can be a little safer, and the use of ChatGPT also allows for the other possible answers and their associated images to be relevant when selecting the correct answer from the offered images, as obviously a test cannot be used where, for example, when selecting the brand of the first own car, next to the picture of the user’s old Polski Fiat there are perfect condition Ferraris and Maseratis displayed.
Thus, artificial intelligence can display additional choices that almost exactly match the parameters of the only correct answer, whether it be pictures of people who resemble the given relatives to a deceptive extent or cars with similar market value even taking into account the level of rust on body parts.
And last but not least, with the ChatGPT module integrated into password managers, it may become possible for the user to continuously search for traces of leakage of their passwords and other personal data on the internet in case of unlimited internet access. Although cybersecurity companies behind password manager software themselves constantly monitor users’ authentication data, with ChatGPT capabilities, the user can also use their own computing power for this purpose.
This continuous search, adapting to the paranoia of the given user, can be of individual depth for every device on which the password manager is installed, covering a wide range from maximally energy-saving to not sparing the electricity bill.
(I wrote this concept for an article for the Future Factory website, where it was published in Hungarian on March 22, 2023. The English translation for this blog was made with ChatGPT, so apologies for any possible grammatical errors.)