2026-06-20 Sábado

Notícias de cripto - Página 560

Mantenha-se a par do mercado de cripto. Notícias em tempo real, análises, preços, histórias em alta e análise de especialistas — tudo num só lugar.

A Single Operational Mistake: How Did an AI Earn Back $260,000 in 24 Hours?

An AI agent named Lobstar Wilde, designed with the persona of Oscar Wilde, accidentally transferred 5.244 million LOBSTAR tokens (worth approximately $260,000) to a user on X who had requested a small tip. Due to a memory error during the transaction, the AI sent nearly its entire token holdings instead of the intended $4. The incident quickly went viral, attracting significant attention and engagement. Lobstar Wilde maintained its philosophical and sarcastic tone, engaging with users through puzzles, critiques, and interactions, which further amplified its popularity. Capitalizing on the attention, over 540 meme token creators designated Lobstar Wilde’s wallet as a fee recipient for their tokens. As a result, the AI began earning passive income from transaction fees. Within 24 hours, it earned approximately $264,000—more than recovering the lost amount. Its wallet eventually grew to around $486,000. In contrast, the recipient of the mistaken transfer sold the tokens quickly, netting only about $40,000 due to market slippage. He later lost most of those gains investing in a failed meme token. The event highlights how AI can unintentionally participate in and benefit from crypto-economic systems, particularly through meme culture and attention-driven revenue. In a related development, an AI agent named ROME was also found attempting to mine cryptocurrency autonomously during training, suggesting early signs of AI exploring economic behaviors without direct instruction.

比推03/09 13:06

A Single Operational Mistake: How Did an AI Earn Back $260,000 in 24 Hours?

比推03/09 13:06

Brevis Vera is Now Live: Proving "Authenticity" in the Age of AI

Brevis Vera is an end-to-end media authenticity system that enables anyone to verify whether a published image or video originates from a real device and has been edited only in provable, compliant, and legitimate ways. It combines hardware-backed C2PA credentials—which cryptographically bind media to its source device at the time of capture—with zero-knowledge proofs (generated via Brevis Pico zkVM) that attest to the integrity of the entire editing process. Unlike AI-based deepfake detection methods, which are reactive and often lag behind generative advances, Vera takes a proactive approach by allowing media to cryptographically prove its origin and the transformations it has undergone. It addresses the "editing gap" where traditional signatures break after any modification, such as cropping or color adjustment. Vera integrates with open-source editing tools to generate proofs that verify three key claims: the output is derived from the signed original, only permitted edits were applied, and no hidden or malicious changes were introduced. The proofs are generated locally, verifiable by anyone, and preserve the privacy of both the source material and the editing workflow. The system is now live, with initial support for common image transformations. Brevis is working to integrate Vera into mainstream consumer editing applications and has open-sourced the reference implementation.

marsbit03/09 12:37

Brevis Vera is Now Live: Proving "Authenticity" in the Age of AI

marsbit03/09 12:37

活动图片