Cryptocurrency exchanges Binance and Coinbase are in discussions with AI startup Anthropic regarding access to the Claude Mythos Preview model, according to The Information.
The project released a new neural network in early April, but it is not publicly available. Participants in Project Glasswing, including AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, and others, will be able to test Mythos in secure conditions.
The restrictions are due to the model’s reported capabilities. Developers claim it can scan millions of lines of code and identify vulnerabilities in seconds. During tests, the neural network discovered a 27-year-old vulnerability in OpenBSD and a 16-year-old bug in FFmpeg.
Anthropic’s research indicated that AI agents based on Mythos could autonomously hack smart contracts, causing simulated damage worth millions of dollars.
Traditional audits take weeks, whereas the new model finds effective attack paths in seconds.
“Mythos and future models will enable even deeper testing of software and systems at scale. This will accelerate both digital threats and digital defenses,” said Coinbase’s Chief Security Officer Philip Martin, confirming the negotiations with Anthropic.
Binance’s Head of Security Jimmy Su also emphasized that the exchange is exploring “how AI advancements can bolster cybersecurity while simultaneously creating new risks.” He noted that the platform is experimenting with AI to more swiftly identify vulnerabilities in internal systems.
Clear Street analyst Owen Lau noted that the threat posed by Mythos is not yet fully understood. He believes that for cryptocurrency exchanges, this will become a negative narrative, but only temporarily.
“In the long term, these platforms will be among the first to defend against such AI agents,” he remarked.
Regulators worldwide have expressed concern over Mythos’s capabilities. In the United Kingdom, emergency talks are already underway with the government cybersecurity agency and major banks.