BREAKING: $435 million has been liquidated from the cryptocurrency market in the past 12 hours.

On May 7th, per Lookonchain monitoring, HyperLabs unstaked 421,879 HYPE tokens two days earlier—valued at roughly $18.08 million—and deposited 400,000 HYPE tokens into crypto exchanges Bybit and OKX over the past 11 hours, worth around $17.34 million.

On May 7, PeckShield monitoring revealed that crypto market maker TrustedVolumes was hacked, resulting in approximately $5.9 million in asset losses. The stolen funds include roughly $3.02 million in ETH, $1.37 million in WBTC, and $1.47 million in stablecoins. The attacker has since converted the stolen assets into around 2,513 ETH. TrustedVolumes was previously hacked in 2025 due to a vulnerability in the old smart contract of 1inch Fusion v1, losing about $4.5 million (primarily USDC and WETH). After negotiations, the hacker returned most of the funds and received a 10% "bounty" reward. That incident did not impact end users, only affecting third-party market makers.

On May 7, Tencent announced that since its launch, the token usage of the Hy3 preview has continued to grow, now totaling more than 10 times that of its predecessor, Hy2. Notably, token invocation has surged in code and intelligent agent scenarios—rising over 16.5 times in Tencent’s WorkBuddy/CodeBuddy and Qclaw-like applications. ### Key adjustments for American English conventions: 1. **Date format**: Uses "Month Day, Year" (standard in U.S. media). 2. **Conciseness**: Replaces wordy phrases ("previous generation model") with concise terms ("predecessor"); simplifies "Token invocation volume" to "token usage" (more natural in tech context). 3. **Emphasis**: Uses "surge" and an em dash to highlight key growth in niche scenarios (aligns with U.S. news brevity). 4. **Terminology consistency**: Keeps product names as given, with consistent capitalization (CodeBuddy instead of codebuddy for brand consistency). All original information is retained without extra characters.

On May 7, cryptocurrency exchange Bithumb announced it would list EDGEX on its Korean won (KRW) market. Following the listing announcement, EDGEX’s price surged more than 6% in a short window, pushing its market capitalization above $500 million.

**May 7th** Andreessen Horowitz (a16z) has doubled down on AI and crypto with a flurry of recent moves: it launched a $2.2 billion Crypto Fund 5 and is ramping up bets on AI agents, enterprise automation, and AI content creation. On May 5, a16z announced its fifth crypto fund—$2.2 billion Crypto Fund 5—to invest in blockchain startups across all stages over the next decade. The firm noted that while some VCs are shifting focus to AI, the crypto industry is at a “historical high,” with priorities on stablecoins, payments, DeFi, prediction markets, and real-world assets (RWA). Simultaneously, a16z is accelerating AI application-layer investments: - **Enterprise AI**: Led Series A funding for Tessera Labs, an AI system integration firm that uses a multi-agent system to replace traditional SAP consulting teams. The company claims its 6-person AI team can complete the work of roughly 60 consultants in two years. - **AI Talent Platform**: Led a $22.75 million Series A round for L

**a16z Leads Seed Round in Glif, AI’s "Creative Super Agent" Platform** In an official April 23 announcement (made public May 7), venture firm a16z revealed it led the seed investment in Glif—an AI creation platform positioned as a "super AI agent for the creative field." Glif aims to unify generative AI tools (images, video, audio, music composition, voiceovers) into one interface, letting users complete end-to-end creative workflows via natural language. The team notes today’s AI creation ecosystem is overly fragmented, forcing users to switch between multiple products. Glif wants to unify disparate models into a single agent—drawing a parallel to "Claude Code for creative work." Its core feature: "Skills," standardized workflows built from millions of AI creation experiences. These support popular use cases (short videos, AI avatars, infographics, music videos) and let users train personalized styles and processes. **Founder Backgrounds:** - Fabian Stelzer: Founded AI