A GitHub account under the name rabbitscam recently posted the source code of Rabbit R1, an AI assistant device showcased at CES earlier this year, claiming that there is no actual AI technology within it. In reality, the capabilities of Rabbit R1 are limited to a Playwright script that controls the browser.
However, a screenshot of a Discord message from Peiyuan Liao, the CTO of Rabbit Inc., on Hacker’s News reveals that the Large Action Model (LAM) operates outside of the code that rabbitscam posted. It is not uncommon to utilize external scripts for command execution, as agents running on the LLM system often involve additional layers of scripts. Nevertheless, the implementation strategy of Rabbit R1 is quite risky, as it directly requests user passwords without controlling various apps through APIs.
Another product similar to Rabbit R1 is the Humane AI Pin, which was reviewed by MKBHD as one of the worst products ever reviewed.
Source: GitHub: rabbitscam, Hacker’s News
TLDR: The GitHub account rabbitscam posted Rabbit R1’s source code, revealing limited AI capabilities, while the Humane AI Pin was dubbed one of the worst products reviewed.
Leave a Comment