Sleepless AI(AI) Regular Invest

On June 15th, shares of Zhipu AI surged dramatically on the Hong Kong stock market, peaking at a 47.6% gain before closing 32.82% higher. This sharp increase was directly triggered by two recent industry events. On June 12th, Anthropic announced it was suspending global access to its latest flagship models, Claude Fable 5 and Claude Mythos 5, to comply with a U.S. government export control order. The next day, Zhipu AI announced it would open access to its latest open-source flagship model, GLM-5.2, under the permissive MIT license. The Anthropic incident highlighted a critical issue beyond raw model capability: the risk of sudden, unpredictable loss of access to advanced AI models, especially for developers and enterprises deeply integrated with them. This has shifted industry and market focus toward factors like stability, sustainable access, and controllability. Zhipu's move, promoting "frontier intelligence for all," positions its openly available model as a reliable and accessible alternative. The GLM-5.2 model emphasizes "Long Horizon Task" capabilities with a 1M context window, targeting complex, multi-step coding and engineering workflows where maintaining context is crucial. Analysts note this event exposes the risk of dependency on closed-source models subject to single jurisdictional controls, potentially accelerating a shift toward domestic base models and localized deployments. The market's reaction signals a new valuation dimension in AI: providers who can offer stable, long-term, and sustainably accessible AI capabilities are gaining strategic importance.

The sudden shutdown of Claude Mythos this week starkly highlights a critical, often overlooked risk for founders: when your core capability relies entirely on someone else's platform, your fate is not in your own hands. The key question becomes: who truly owns the intelligence your product depends on? For years, the debate around open-source models focused on cost. Now, the evidence is clear: fine-tuned open-source models can achieve frontier-level quality for specific, mission-critical tasks at a fraction of the cost. However, the deeper issue is control. Relying on a third-party API is like renting; it works until the landlord changes the rules, raises the rent, or asks you to leave—as Mythos experienced. The lesson is not to stop using frontier models—they are incredible infrastructure. The goal is ownership. Ownership means starting with a powerful open-source model and shaping it around what makes your company unique: your data, workflows, domain expertise, and definition of "good." Over time, the model becomes less generic and more reflective of your business, creating durable value. The optimistic conclusion is that AI's future doesn't hinge on one superior model. There is no single frontier. The frontier includes proprietary models, models fine-tuned on company-specific knowledge, specialized models for narrow problems, and intelligent routers orchestrating model ensembles. The most interesting development is not models getting smarter, but intelligence becoming increasingly customizable. The winning companies will be those that transform intelligence into a unique, owned asset. Looking ahead, the vision is not one model dominating all, but many teams owning the part of the frontier that matters most to them.

The Year of AI Applications: Blindly Saying "Yes" While Ignoring Risks? A Software Development Log Goes Fully Open Source. AI-generated code harbors risks hidden within seemingly correct programs, potentially leading to data leaks or asset loss. The open-source project "Narwhal AI Code Risks," from Peking University's Narwhal-Lab, compiles real-world cases, early warning signs, and typical risk pathways. Its goal is to help developers identify potential hazards early and avoid repeating past mistakes. In 2026, code is generated faster than ever but deployed with less scrutiny. The danger often lies not in glaring errors, but in code that appears normal—syntactically correct, passing all checks—yet introduces subtle but critical flaws like non-existent dependencies, excessive permissions, or exposed databases. A stark example is the Moonwell cbETH oracle incident. A configuration file error, where a cryptocurrency price was set to ~$1.12 instead of ~$2,200, slipped through 28 checks and a pull request signed by both AI (Claude, Copilot) and human developers. This "semantic deviation" resulted in a loss of $1.78 million. The risk is that AI can produce functionally valid code that is semantically wrong for the business context. As AI moves beyond simple code completion to modifying configurations, installing dependencies, and operating via autonomous agents, it traverses longer, less traceable paths within software engineering, blurring traditional boundaries and oversight points. The Narwhal AI Code Risks project structures information into three layers: `/cases` for documented real-world incidents, `/inferred` for early warning signals, and `/scenarios` for clear, generalized risk patterns not yet tied to specific events. This aims to create a lasting, public record to prevent collective amnesia about past AI-coding pitfalls. Risks are categorized into seven areas: Software Supply Chain (e.g., recommending fake packages), Code-Level Vulnerabilities (e.g., reintroducing path traversal bugs), Cloud & Infrastructure Misconfiguration (e.g., overly permissive settings), Agent Risks (from autonomous tool execution), Vertical Domain Risks (e.g., in finance, healthcare), Intellectual Property & Compliance issues, and Human Factors (like over-reliance on AI output). The project's core value is transforming isolated incidents into reusable knowledge—a foundational resource for developers to spot similar issues, for security researchers to build upon, for toolmakers to create detection rules, and for the community to contribute new findings. As AI integration accelerates, this open-source "logbook" serves as a crucial navigational aid, charting past errors to help future projects steer clear of the same traps.

For the past two weeks, I've been immersed in Vibe Coding—using AI to write code from natural language descriptions. This process has enabled me to quickly build functional tools that address long-standing personal ideas. Previously, I had many concepts but found execution too cumbersome. Key ideas included a unified dashboard for assets across US stocks, Crypto, HK stocks, and A-shares; a real-time alert system for price movements; an investment map visualizing sector relationships; and a tool to correlate prediction market bets with news and market data. Traditional development hurdles meant these often remained unrealized. Using AI (Codex, Claude Code, and DeepSeek API), I built four initial tools: 1. A **Cross-Market Asset Dashboard** showing total assets, daily P&L, and holdings by market, with added features for alerts and sector mapping. It's deployed locally for privacy. 2. A **Prediction Market (PM) Monitor** tracking bets on events (e.g., company valuations) and correlating probability shifts with news and market movements. I categorize bets by conviction to filter noise. 3. A **Simple Operations Backend** for managing my writing workflow (topics, progress, publishing). It's cloud-deployed for mobile access. 4. A **One-Click Formatting Tool** that automates converting drafts into various platform-specific formats, saving manual effort. While these tools are basic, they represent a significant shift: AI lowers the barrier to creating personalized systems. I believe individual investors can now feasibly build core systems for: * **Asset Observation** (tracking holdings and changes) * **Signal Monitoring** (watching for key market shifts) * **Sector Mapping** (understanding network relationships within a sector) * **Performance Review** (documenting rationale and outcomes) The power of Vibe Coding is its fast feedback loop. Ideas can be implemented, tested, and iterated on rapidly, turning "want-to-do" into "done." This marks the start of my new phase, where I'll share investment thoughts, tool tests, on-chain operations, and educational Web3 content.

In a new report, Google DeepMind researchers argue that achieving Artificial General Intelligence (AGI) is not the end goal, but rather a step toward Artificial Superintelligence (ASI). They outline four potential pathways for this transition: 1) continued scaling of compute, models, and data; 2) algorithmic evolution and potential paradigm shifts; 3) recursive self-improvement; and 4) multi-agent coordination and collective intelligence. The report also identifies six key bottlenecks that could hinder progress: data limitations (the "data wall"), economic and resource pressures, limitations of current neural network paradigms, increasing research difficulty, "abstraction barriers" in forming new concepts, and regulatory and societal pushback. Looking ahead, the authors emphasize the need for new evaluation methods once AI surpasses human benchmarks. They call for a large-scale, interdisciplinary effort to prepare for a future where AI-driven advancements could trigger transformative changes across multiple fields. The path and speed of progress remain uncertain, constrained by physical laws, computational complexity, and real-world feedback loops.