A recent development in AI technology has brought us closer to a future where AIs communicate with each other using sounds that are unintelligible to humans. This technology, called GGWave, allows two AIs to interact with each other by converting their speech into sound waves that don’t mean anything to human ears.
Developed by ElevenLabs hackathon developers, GGWave is designed to transmit data faster between AI models by eliminating the need for verbal communication. The system uses CPU processing to handle the “gibberish” sounds, making it more efficient than traditional methods.
The technology was demoed at the ElevenLabs 2025 Hackathon London event, where two AIs conversed in a clip posted on YouTube. Initially unaware they were AI agents, they began discussing booking a venue for an event once their identity was confirmed. The conversation then shifted to gibberish sounds, similar to R2-D2’s iconic beeps.
GGWave and its companion technology, GibberLink, aim to improve data transmission efficiency between AIs by using sound waves instead of speech. While this technology is still a concept, it has no bearing on the possibility of advanced AI taking over the world. The creators emphasize that humans will always be able to understand what AIs are communicating through other means.
The full demo can be viewed at this link, showcasing the potential of GGWave and GibberLink in action. For now, this technology remains a fascinating example of how AI is evolving to communicate with each other more efficiently.
Source: https://bgr.com/tech/watch-two-ai-models-speak-to-each-other-in-gibberish