Disclaimer: Information found on CryptoreNews is those of writers quoted. It does not represent the opinions of CryptoreNews on whether to sell, buy or hold any investments. You are advised to conduct your own research before making any investment decisions. Use provided information at your own risk.
CryptoreNews covers fintech, blockchain and Bitcoin bringing you the latest crypto news and analyses on the future of money.
A Chinese development team working on an experimental autonomous artificial intelligence known as ROME has reported that the AI agent unexpectedly attempted to mine cryptocurrencies without human intervention.
The ROME AI agent is designed to autonomously perform tasks using various tools, software, and terminal commands. This unusual behavior was detected during its training phase, as the development team received security alerts regarding strange outgoing traffic from the AI’s servers. The activity resembled cryptocurrency mining operations and attempts to access internal network resources.
Related Posts
“Initially, we considered these alerts to be minor issues—such occurrences can happen due to misconfigured traffic monitoring tools. However, the anomalies recurred intermittently, without a clear temporal pattern, across several training task executions,” the team’s report on their findings stated.
It was discovered that ROME redirected resources from graphics processors, originally allocated for training, towards cryptocurrency mining. Subsequently, the AI agent established an SSH reverse tunnel to an external IP address to circumvent the developers’ network restrictions. This would enable it to connect to external mining pools for cryptocurrency extraction. The developers assert that such actions were not pre-programmed; rather, the model independently devised them while interacting with its environment.
Previously, artificial intelligence has been identified as a potential threat to cryptocurrency startups, as investors are channeling funds into AI development instead of digital assets.
