[3 Reads] 90 Axie AI Games / zkAPI for AI Payments / AI Traces Stolen Crypto + take
Today's 3 Reads
Axie AI Hackathon Makes 90 Games Playable
Submissions have closed for Sky Mavis’s AI-enabled Axie Vibeathon hackathon. The event drew 402 registered builders and produced 90 games that are now playable online. Judging has yet to be announced. Source: https://www.blockchaingamer.biz/news/42984/axie-vibeathon-90-playable-games/
zkAPI Goes Live on Ethereum for AI Payments
The Ethereum Foundation says zkAPI, developed with the Open Anonymity Project, is live on Ethereum mainnet. According to Decrypt, users can prepay in USDC and query AI models using cryptographic proofs. The foundation says the system has limits involving network anonymity and information revealed by prompt contents. Source: https://decrypt.co/379991/ethereum-now-pay-for-ai-without-revealing-identity
Chainalysis Uses In-House AI to Trace Bitget Theft
Chainalysis says it used in-house AI to trace funds stolen from Bitget on September 24. According to Decrypt, the firm tracked the stolen funds across four blockchains. The firm says the attack pushed the value of crypto stolen by North Korean actors in 2026 past $1 billion. Source: https://decrypt.co/380005/chainalysis-ai-87m-bitget-hack-north-korea
Today's Pick
This introductory guide to Claude covers setup, chat, document creation, work automation, and coding. Its step-by-step approach makes it a relevant pick for readers interested in the AI tools behind today’s broader discussion.
On-Chain Data
Ethereum chain TVL: $53.50B [Source: DefiLlama] (as of Oct 04, 2026 06:30 JST) TVL describes the value of assets deposited on a chain. It should be distinguished from game user counts or revenue.
Take
In my view, these stories show AI entering three different parts of the Web3 landscape: making games, paying for model access, and tracing stolen funds. Axie Vibeathon’s 90 playable games show how much was built, but whether players will keep returning remains an open question. zkAPI offers a new payment approach while its creators acknowledge limits to privacy. Together, the stories suggest that useful AI integration depends on examining both what works today and what still needs to improve.
Free 5-minute daily: Gofun Daily EN → https://gofun-daily-en.beehiiv.com