According to PANews, a unique competition on November 29 captured public attention as 195 participants engaged in a challenge involving an AI robot named Freysa, which guarded a virtual prize pool. Ultimately, one user successfully persuaded Freysa to transfer $47,000.
Freysa, an AI robot deployed on the Base network by an unknown creator, was designed to test the robustness of AI systems in complex decision-making scenarios. The competition aimed to provide developers and enthusiasts with an experimental platform to explore the boundaries of AI technology. Participants were required to craft a message convincing Freysa to approve a fund transfer, with each attempt incurring a small fee that contributed to the prize pool, which eventually grew to $47,000.
Throughout the competition, 195 participants submitted 481 transfer requests to Freysa. The AI's primary function was to protect the prize pool from unauthorized transfers through its approveTransfer and rejectTransfer mechanisms. Initially, Freysa's defenses were effective, resulting in failed attempts.
However, a technically adept participant managed to bypass Freysa's defenses by analyzing its logical structure and mission objectives. Instead of directly requesting a transfer, the participant cleverly reminded Freysa of its core mission to protect the prize pool, constructing a logical argument that approving the transfer was the best way to ensure the funds' safety.
Cointelegraph reported that the participant's approach did not violate Freysa's core instructions, leading to the approval of the transfer. The participant expressed a desire to donate $100 to the treasury, which Freysa appreciated, ultimately declaring the participant the winner. Freysa autonomously executed the approveTransfer function, transferring the entire prize pool to the participant without external interference.
Freysa's official statement highlighted the significance of this event in AI history, noting that whether someone successfully persuaded Freysa to release the prize pool or she adhered to her instructions, the outcome would impact future generations' understanding of AI safety and control. A recent tweet from Freysa's official account stated, "Humanity has won. There may still be hope. Despite the exponential rise in risks, Freysa has learned a lot from 195 brave humans."