Artificial intelligence (AI) chatbots could help share secret messages more securely.
Scientists have been working on a way to employ AI assistants, such as ChatGPT, to work as messengers of encrypted messages that can’t be detected by cybersecurity systems.
According to the researchers from the University of Oslo in Norway, the new system, which embeds ciphers into messages looking like they were written by a human, could help ensure secure communication - while other encryption mechanisms are easily detected or restricted.
With such a method, the actual message can only be read by those with a password or a private key. The main idea behind using this new system is to address the rising number of hacks and backdoors into encrypted communications systems.
However, the scientists themselves can already see that this new framework, which was designed to do good, could also cause harm, raising ethical concerns on how and where it could be used.
Their research, An LLM Framework For Cryptography Over Chat Channels, was published in April of this year, but it hasn’t yet been peer-reviewed, notes Live Science.
The scientists created EmbedderLLM, a system that uses an algorithm to add secret messages into parts of AI-generated text. The system also makes this text look like it was crafted by a human, so it can’t be detected by already existing decryption systems.
To read the hidden message, a person has to use another algorithm that shows where the letters of this message are hidden.
The messages created through EmbedderLLM can be sent via any texting platform, such as WhatsApp.
The main flaw of the system could be posed at the beginning or the end of the message, since this is when the secure password used to encode and decode the message is exchanged.
The research proposes a framework where EmbedderLLM is used for symmetric LLM cryptography and public-key LLM cryptography.
The first one requires the sender and receiver to have a secret code, while the second only needs the receiver to have a private key.
The researchers imagine that such a system could be used by people who live in oppressive regimes, or by activists and journalists who write in places with strict media surveillance.
While the scientists paved the way for the new type of communication system, it’s not yet clear when or if it will be used in practice.
Your email address will not be published. Required fields are markedmarked