For many years now, we have used human data to train AI software to become more and more human-like. For example, we build emotive interfaces to please us; while this is positive for the user experience and for improved interactions between humans and machines, it opened the door to real dangers. We rely on CAPTCHA to distinguish between humans and bots. However, researchers from the University of Maryland can fool Google's reCaptcha using Google's own speech-to-text service with 90% accuracy. So, the question becomes: what would prevent AI from using all of our information to create what is called “deepfakes”? The way to eliminate this possibility is to adopt liveliness as a standard for online activities. Physiological biometrics unlocks a foolproof, frictionless way to ensure a live human is operating at the other end of a connection. We offer Touch Technology®, which uses a live signal to determine a human from a bot, and A-Live®, which can detect a live signal when a person interacts with a device by holding it in the palm of their hand.

Deep fakes are computer generated personas that resemble a person but are fake. They are so realistic that they cannot be distinguished from the real person. Computer Graphic Imaging (CGI) can create ‘real’ fake actors, for example what was done in the latest Star Wars movie. We have been using CAPTCHA to segregate humans from bots for years now, but we also used all of that data to train AI to read CAPTCHAs. Now it is obsolete, and CAPTCHA2 came to replace it with mosaic imaging, but the same thing happened. So, we have a cycle: we build a tool to differentiate bots from us, then we train the bots to be us.

Physiology contains vital signs that are never the same twice, making it a superior option for liveness detection, and a key component needed to strengthen existing biometry against spoofing attacks. For this reason, Aerendir Mobile developed A-Live®, which uses physiology to determine the presence of a human, and Touch Technology® which uses physiology to distinguish non-human activity from human activity, thereby making an anti-bot solution.

A-Live® uses physiology to detect the true presence of a human holding the device. Precisely, a pre-trained AI uses the live physiological signals originating from the nervous system to determine the presence of a human. A-Live® identifies human nervous system data using the existing sensors in a mobile device and is only triggered by any third party app or user. It also consolidates every other biometric technology, eliminating the threat of a systemic breach and reinforcing existing security.

Touch Technology® is our anti-bot solution which extracts a unique proprioceptive signature, detected by accelerometers and gyroscopes to safely and effectively recognize if the signal is generated by a human or a bot. Capturing live signals from the individual’s unique NeuroPrint® creates Identity Assurance across the Internet.

The brain and nervous system are very complex, and to be able to generate an acceptable fake signal, one would need to rely on such complexity. So, while this is possible, it is technically and conceptually difficult. Then again, code may be slightly altered to make this training irrelevant.

No, the anti-bot is ready as it is. It does not need user-specific training. It is essentially a business-to-business solution under the shape of a plug-in that can fit inside any mobile app.

Social media platforms of course, but also news editors and entities that need to decrease bot amplification of web content. Once blockchain becomes ubiquitous, there will be a huge need to ascertain that the other party is indeed a bona fide human and not some piece of software impersonating a human. The trust protocol offered by blockchain may be breached if we do not have such a powerful tool. But in the long term, this product – if successful - may make the Internet real again.