By this point, ChatGPT has become a household name as the pioneering example of a novel form of artificial intelligence known as generative AI, or large language models (LLMs). Since its launch in November 2022, ChatGPT has experienced exceptional growth, amassing a staggering 1 million users in just five days.
Today, with over 100 million users, ChatGPT has become a widely adopted AI model, and Openai.com, the platform behind its creation, receives approximately 1 billion monthly visits.
As our reliance on LLMs continues to grow, our digital footprints expand exponentially.
This is due to our usage of LLMs, like other websites and applications, through internet and satellite communications that are safeguarded by public key infrastructure (PKI).
How quantum computers are a threat to Langauge Models?
In recent times, quantum computers have gained significant attention for their immense computational power. These machines have the potential to break PKI and decrypt vast amounts of our existing communications, encompassing almost everything we engage in on the internet. Consequently, the data we generate, including LLM data, becomes vulnerable to exploitation by adversaries equipped with quantum computers.
Quantum threats arise due to the development and potential utilization of quantum computers, which possess the ability to perform certain calculations at a much faster pace than classical computers. These highly capable machines pose a significant risk to the security of current cryptographic algorithms used for safeguarding data.
A primary concern is the susceptibility of traditional asymmetric encryption algorithms, such as RSA and elliptic curve cryptography (ECC), to quantum computers. These algorithms rely on complex mathematical problems, like factoring large numbers or solving discrete logarithms, to ensure the integrity of encryption. However, quantum computers can efficiently solve these problems using algorithms like Shor's algorithm, potentially compromising the security of encrypted data.
Why it is very challenging to protect a system from quantum threats?
As we increasingly engage with technology in our daily lives, such as using mobile phones, making online purchases, or utilizing AI-powered systems like LLMs, our digital presence expands. This accumulation of data creates a digital twin or shadow that grows alongside our activities.
With the advancement of LLMs and AI, it becomes increasingly challenging to differentiate between our real selves and our digital counterparts.
The integration of LLM data into our digital shadows means that every aspect of our lives becomes part of a permanent and accessible database. This poses a potential risk as adversaries equipped with quantum tools could gain unauthorized access to this data.
Consequently, not only individuals but also businesses and government organizations that rely on these technologies expose themselves to the same threats.
Consider the gravity of sensitive information such as government or nuclear secrets, personal healthcare records, bank account details, and identification numbers like Social Security and driver's license numbers. It is imperative that these data remain confidential and secure for extended periods, ranging from 25 to 75 years.
However, if a highly capable quantum computer becomes operational within the next five years, the potential consequences become evident. The exposure of our most valuable and sensitive information would persist for decades, compromising privacy and security on an unprecedented scale.
What you can do to protect your data privacy in the era of LLMs and AI technology?
- Understand the implications: Recognize that the use of LLMs and AI exposes data during communication and storage processes. Be aware of the potential risks associated with these technologies.
- Avoid sensitive data exposure: Refrain from using LLMs or AI platforms where there is a possibility of sensitive or confidential information being exposed or stored. Exercise caution and prioritize data privacy.
- Review privacy policies: Take the time to read and understand the privacy policies of LLM websites or AI service providers. Being aware of how your data will be used can help you make informed decisions about sharing information.
In response to this challenge, researchers are actively engaged in the development of post-quantum cryptography, also known as quantum-resistant cryptography. This field focuses on creating new cryptographic algorithms that can withstand attacks from both classical and quantum computers.
The ultimate objective is to establish long-term security for sensitive data and communications, considering the increasing power and accessibility of quantum computers.
Ongoing efforts are underway to standardize quantum-resistant cryptographic algorithms and establish a new generation of encryption methods.
These endeavors involve collaboration among experts to define and promote secure algorithms capable of countering quantum threats. Furthermore, organizations may need to adopt hybrid approaches that combine classical and post-quantum cryptography to ensure the secure transmission and storage of data in the face of evolving quantum threats.