New ARS research by t54 Labs and collaborators from Google DeepMind and Microsoft aims to standardize risk with escrow and collateral systems for agent-based economies

61% loss reduction in simulations sounds great until you see that zero-loading premiums left underwriters insolvent — same unsolved pricing problem DeFi insurance has been stuck on since Cover Protocol collapsed. Collateral requirements alone deterred 15-20% of risky transactions, which tracks with how margin works in TradFi: skin in the game filters bad actors before insurance even enters the picture. Virtuals Protocol co-authoring this alongside DeepMind and Microsoft puts the biggest AI agent token launchpad at the same table as the labs building the models — whoever controls the settlement layer for agent-to-agent transactions controls the value capture.

Top comment by @Benthic

More coverage

• New standard aims to make AI agents financially safe - Global Crypto (globalcrypto.tv)
• New Standard Aims to Make AI Agent Spending Safe | Krasa.ai (krasa.ai)
• Researchers Propose Financial Risk Standard for AI Agents · Blockster (blockster.com)
• Microsoft Research AI Frontiers (microsoft.com)
• DeepMind - AI Lab Watch (ailabwatch.org)

Explore the topic

• Research
• Microsoft
• Google
• AI Agents