Hand-picked deep dives and narrative features from the BlockAI News desk.
Editor Pick
Microsoft's six-year OpenAI exclusivity ended April 27. By April 28, OpenAI was on AWS — with a $138B total compute commitment. Inside the deal that rewrote AI's cloud map.
Editor Pick
Beijing's NDRC ordered Meta to unwind a $2B acquisition already signed, integrated, and announced. Combined with travel bans on Manus co-founders, the move sets new precedent for cross-border AI M&A and forces every Western acquirer to reprice Chinese-founded AI startup risk.
Security
An explicit 1-of-1 DVN configuration turned a sophisticated RPC infrastructure attack into a $292M systemic asset release event. Full on-chain forensics and root cause analysis.
AI Agents
Cobo just launched an AI-native wallet supporting 80+ blockchains, escalating an agentic wallet race that Coinbase opened in February. Inside the three launches in 70 days that built the production stack for autonomous on-chain AI — and the security gap nobody has solved.
Security
The Kelp DAO exploit didn't stop at Kelp. It left Aave with $196M in bad debt and wiped $6B from its TVL — exposing how liquid restaking tokens can trigger cascading failures across DeFi's most critical lending markets.
Security
A compromised third-party AI tool gave attackers access to Vercel's internal systems and customer environment variables — sending Web3 teams scrambling to rotate API keys and exposing a new class of supply chain risk hiding in plain sight.
Funding
OpenGradient raises $9.5M led by a16z crypto and Coinbase Ventures to build a decentralized compute layer for verifiable AI — where every inference comes with cryptographic proof, auditable on-chain.
Funding
Nava raised $8.3M in seed funding led by Polychain and Archetype to build a blockchain verification layer for AI-powered financial agents — addressing $2.1B in unauthorized AI transactions in 2025.
DeFi
Kelp DAO lost $292M in 2026's biggest DeFi hack after an attacker manipulated LayerZero's cross-chain bridge. Aave, Lido, and Ethena all froze related markets within hours as contagion spread across 20+ chains.
AI
Hermes Agent by Nous Research has hit 97,000 GitHub stars in 8 weeks. Behind the numbers is a Web3-native AI lab backed by Paradigm and a16z — and an unresolved token question the crypto community can't stop talking about.
News
Anthropic's new Claude Opus 4.7 delivers stronger agentic coding, better self-verification, and new cost-control tools — at the same price as Opus 4.6. Here's what it means for Web3 teams running AI agents.