AI Industry Development & Policy Dynamics

After Anthropic’s Claude Code feature was just released on 4/8, the 4/9 update unexpectedly removed it, leaving the developer community severely disappointed. The original electronic pet Buddy sparked a lot of discussion due to leaked source code, but it disappeared within just a day, causing a strong backlash across the community—some people even chose to downgrade and keep an older version. Developers feel betrayed by this and are worried about the future return of the functionality.

Meta is shifting its focus of resources to generative AI, and Zuckerberg is personally using AI tools to write code, showing how much he values development. However, the performance pressure triggered internally by a competition for token consumption could lead to wasted resources and performative behavior. Meta hopes to avoid the failure lessons of the past metaverse era in the AI space, seeking real-world applications and commercial value.

The white paper jointly released by Gate Research Institute and Global HR states that by Q1 2026, the crypto industry will face job shifts driven by AI transformation, and cross-functional talent will become a core competitive advantage. The industry is mainly divided into four areas: exchanges, public chains and infrastructure, DeFi and derivatives, and stablecoins, with technology research and development needs being the most prominent.

Anthropic has launched Claude Managed Agents, providing a fully managed AI agent development platform that simplifies enterprise automation workflows and claims it can accelerate product production by ten times. The platform includes a secure infrastructure, supports long-running execution and multi-agent coordination, and allows developers to focus more on core business operations. Notion has already been an early adopter of this service, improving work efficiency and showing that competition in the enterprise AI market is intensifying.

Anthropic and CoreWeave have reached a multi-year agreement to rent its data center capacity to support growing demand for AI services. The deal will help Anthropic build its Claude AI models and includes multiple Nvidia chip architectures. The partnership will also require working with Broadcom and Google to secure 3.5 gigawatts of power.

April 10, the China Artificial Intelligence Industry Development Alliance held its plenary meeting in Wuhan, emphasizing that artificial intelligence is entering the era of intelligent-native systems, focusing on breakthroughs in model capabilities, the rise of agent frameworks, and the activation of economic growth and security risks through large-scale applications. The alliance will focus on technological innovation and safety governance.

The HY-Embodied-0.5 series of embodied intelligence foundation models released by Tencent includes two versions: 2B and 32B. It focuses on improving spatiotemporal perception and embodied reasoning capabilities, and is suitable for scenarios such as environment prediction and task planning. The 2B model has been open-sourced; its performance is better than other same-scale SOTA models, demonstrating strong intelligent generalization ability and application potential.

Embodied intelligence is reaching a turning point. China’s Zhiren robot company has released GE-Sim 2.0, upgrading world models into a robotic simulator. The technology is designed to address problems such as high training costs and insufficient data, enabling robots to autonomously learn and improve in a virtual environment, and it represents an important advancement for China in robotics technology.

OpenAI has recently faced legal and regulatory crises. The Florida attorney general has launched an investigation into the use of ChatGPT, suspecting it may have helped plan a campus shooting, and it will also face lawsuits brought by victims’ families. In addition, OpenAI is also facing multiple legal lawsuits for encouraging self-harm and generating inappropriate content. The company has said it will fully cooperate with the investigation and has put forward a “Children’s Safety Blueprint” to improve related issues.

Anthropic has launched Claude Managed Agents, providing a fully managed AI agent development platform that simplifies enterprise automation workflows and claims it can speed up product production by ten times. The platform includes a security foundation, supports long-running execution and multi-agent coordination, and helps developers focus more on core business work. Notion has already been the first to apply this service, improving work efficiency and showing that competition in the enterprise AI market is intensifying.

Bittensor co-founder Const responded to community controversy, clarifying token release and management permission issues and emphasizing that his actions are no different from those of ordinary holders. He denied large-scale sell-offs, saying transparency in trading is a core principle, and that community discussions are still ongoing.

On April 10, the authoritative AI benchmarking platform LMArena updated its rankings. GLM-5.1 took first place among global open-source models and third place globally, showcasing excellent coding ability and breakthroughs in long-range tasks. It became the only open-source model to achieve continuous work for 8 hours.

Gate News message: During the 17th plenary session of the China Artificial Intelligence Industry Development Alliance on April 10, the Security Governance Committee of the China Artificial Intelligence Industry Development Alliance officially released the 《OpenClaw Class Intelligent Agent Deployment Risk Management Guidelines》. The guidelines focus on enterprises’ pain points in risk prevention and control during the process of technical implementation, and propose an overall security deployment framework and self-check standards covering the entire lifecycle from deployment, to use, to decommissioning. It aims to help intelligent agents move from “able to deploy” to “secure deployment, standardized use, and ongoing governance,” providing practical guidance for enterprises to carry out technical deployments in a compliant and standardized manner.

Security research reveals that in the LLM agent ecosystem, over 20% of free API routers actively inject malicious code, leading to asset theft and credential crises. In addition, the Claude code-leak incident has enabled attackers to spread malware by exploiting developers’ curiosity. The research team proposes a three-layer defense mechanism to address supply-chain security risks.

Cardano's founder, Charles Hoskinson, highlights the synergy between AI and blockchain, announcing an AI bot for content creation. Cardano's developer group shifts focus to growth, halting Acropolis development to prioritize chain abstraction, while addressing performance issues in their upcoming hard fork upgrade.

Solayer’s founder exposes a security vulnerability in large language model (LLM) routers; in 428 routers, more than 20% exhibit malicious behavior, such as private keys being stolen. The research recommends that developers implement a separate end-to-end integrity verification mechanism on the client side and provides three defense options to mitigate supply-chain attacks.

Anthropic has launched Claude Managed Agents, providing a fully managed AI agent development platform that streamlines enterprise automation workflows and claims it can accelerate product production by 10x. The platform includes a security-focused infrastructure, supports long-running execution and multi-agent coordination, and enables developers to focus more on core business activities. Notion has already been an early adopter of this service, improving work efficiency and signaling that competition in the enterprise AI market is intensifying.

As of March 2026, Alibaba Cloud’s Qwen series models account for more than 50% of global open-source model downloads, with a cumulative total of 942.1 million times. Their smaller-parameter versions are widely favored by developers. With the release of Qwen 2.5 and 3.5, Chinese models have gradually begun to surpass their U.S. counterparts, and the open-source strategy has become a key factor in the AI competition between China and the U.S.

SK Telecom, Arm, and Rebellions sign a trilateral memorandum of understanding to jointly develop AI data center inference server solutions, combining Arm’s AGI CPU and the RebelCard chip to optimize high-density inference and large-scale AI deployments.

Coinbase's developer platform upgrades the x402 protocol, launching the “Upto” plan and enabling a variable pricing mechanism that replaces the previous fixed-fee model to support large language model (LLM) inference and related computing services. This plan is intended to address the issue where users can’t predict costs because they pay fixed fees, with the final amount paid not exceeding the user-set cap. The x402 protocol has been taken over by the Linux Foundation, but transaction volume has continued to decline after the 2025 peak, and it remains to be seen whether the Upto upgrade can boost network activity.