ImageLockDX™ announced, offering security teams a new approach to authentication that mitigates the risks of compromise from malicious actors using artificial intelligence (AI). The ImageLockDX authentication system is designed to become more secure as it grows more capable. Instead of relying on secrets, such as keys or passwords—which are increasingly easy to crack using AI-based tools, ImageLockDX uses a combination of images and uncorrelated passphrases—a method that is effectively impervious to even the strongest brute force threats.
Because there is no relationship between the image and the passphrase, there is no pattern for an AI model to learn. The system produces zero gradient and zero training signal for any machine learning approach to exploit. According to the inventor, the probability of an AI correctly predicting a passphrase from the image alone is approximately 1 in 10⁵⁰—equivalent to random guessing, and mathematically provable.
The result is a credential that cannot be reverse-engineered from personal data, social media scraping, or AI-driven pattern analysis, because there is no correlation to mine in the first place.
Martinez has filed a patent (USPTO App. 19/442,282) on ImageLockDX. The company is positioned for licensing across consumer, enterprise, government, and NATO/EU markets, as well as direct integration by AI platforms seeking authentication that hardens rather than weakens as their own models improve.
Peter Martinez, Inventor of ImageLockDX
Traditional authentication is on a collision course with AI. The smarter the models get, the easier passwords, patterns, and biometric shortcuts become to predict or spoof. We invert that trend. Users select a personal image, a photo that holds real meaning to them, and pair it with a passphrase deliberately unrelated to that image. The system enforces zero statistical correlation between the two.
Peter Martinez, Inventor of ImageLockDX
No one can hack your memories. That's the whole point.
