🚨 A typical AI Agent security incident recently occurred on the Base chain. An attacker sent a carefully crafted Morse code message to @grok, inducing it to output transfer instructions. @bankrbot then directly parsed and executed those instructions, ultimately leading to the transfer of real on-chain assets. Our analysis found that the core issue was NOT that Grok held private keys. Instead, the real problem was: • Untrusted #AI# natural language outputs were treated as executable financial commands • Permission isolation was insufficient • Trust boundaries between AI output and execution systems were poorly defined This incident highlights the growing security risks at the intersection of AI + Crypto Agents.⚠️ Full analysis 👇

EVM UX is still *insane* in 2026 Just lost $6K trying to buy $BOTCOIN (amazing AI crypto project btw) because I pasted the CA to send. @rainbowdotme let me send Base USDC to a token contract with no warning at all. @bankrbot deployed token contract has no ERC20 recovery function. @base could 100% block this at the blockchain layer (could be a UX differentiator) The money is permanently stuck. This is not consumer-ready.

A huge thank you to everyone who joined us for our soiree at #Consensus2025# last week. 🙏 Grateful to have hosted alongside WisdomTree, Persona, and AMCrypto. Until next time. 🌴

Sui + @Talus_Labs = the infra stack making Agent vs. Agent (AvA) gaming a reality. @doppelgamesco is the proof. Join @0xgboost_, @0xgmike, and @ai_product_man as they discuss the convergence of AI + Crypto. 📅 13 May | 3pm PT Set a reminder:

Just bought a bunch of $MiroShark and $Aeon The first one is an open-source multi-agent simulation engine designed to turn documents and raw information into knowledge graphs, then generate thousands of personalized AI agents capable of simulating user behavior across social platforms and prediction markets. Think of it as a large-scale collective intelligence system. The applications are actually massive: • sentiment analysis • crisis/stress testing • synthetic data generation • market behavior simulation • autonomous decision modeling What makes the architecture interesting is the relationship between the two: Aeon acts as the execution layer (automation, operations, monitoring, promotion, self-iteration). MiroShark acts as the intelligence layer (agent simulation, collective reasoning, market prediction, complex decision analysis). One functions as the operating framework The other functions as the simulated “brain.” Together they form a self-reinforcing autonomous AI system. The dev behind it, Aaron Mars (@aaronjmars), is a Toronto-based AI researcher focused on the intersection of AI + crypto, with multiple open-source projects already shipped including MiroShark, Aeon Agent, and the AI personality framework soul.md. The GitHub activity is active, commits are frequent, and the entire stack is being built in public. Feels like one of those projects people dismiss early because it sounds too technical — until the market realizes it’s building actual AI coordination infrastructure instead of another chatbot wrapper.