AI agents are getting better at managing wallets and executing trades, but I'm still a bit nervous about giving them access to real money.

Remember the Owockibot incident earlier this year? Its wallet keys were reportedly exposed, despite instructions to keep them private.

I've been reading about how @Acurast uses hardware-backed TEEs on smartphones to isolate sensitive code and data, even from the device owner.

It wouldn't prevent every kind of mistake an AI agent could make, but reducing who can access sensitive data seems like a good place to start.

I'd definitely want to know how an agent protects its keys before trusting it with my wallet.