- The AI Report
- Posts
- π Anthropic Models Go Global + Tencent's Hy3 Outshines GLM-5.2
π Anthropic Models Go Global + Tencent's Hy3 Outshines GLM-5.2
Plus: Google DeepMind's creative partnership, OpenAI's regulatory shift, and Alibaba's employee ban on Claude Code
Anthropic's AI models are now set for global release, signaling a new era in AI deployment.
The US government's decision to lift export curbs on Anthropic's models reflects a cautious optimism towards AI's potential, paving the way for innovative public-private partnerships.
Plus, Tencent's Hy3 model takes the lead in open-source AI, while Google DeepMind forges a groundbreaking partnership with A24.
HR and IT need to work as one. Here's how
Onboarding, offboarding, role changes, leaveβevery employee lifecycle moment requires HR and IT to move together. When they don't, people fall through the cracks. Access delays mount. Compliance risk creeps in.
This guide gives HR and IT leaders a practical communication framework to close the gaps, standardize handoffs, and keep the employee experience seamless from day one to last day. Free downloadβbuilt for ops teams that need it to actually work.
π After spooking Trump into safety testing, Anthropic AI models get global release

- Export curbs lifted - The US government has removed export restrictions on Anthropic's Fable 5 and Mythos 5 models, allowing for global access and signaling a shift in regulatory stance towards AI technologies.
- Enhanced safety measures - Anthropic has implemented stronger safeguards in Fable 5 to prevent potential misuse, although this has resulted in some benign prompts being blocked during routine coding tasks.
- Public-private collaboration - Anthropic is deepening its partnership with the government, establishing a framework for assessing AI jailbreak risks and enhancing the rapid response to emerging threats.
The Bigger Picture: The lifting of export curbs on Anthropic's models signifies a cautious yet optimistic approach by the US government towards AI deployment, reflecting a growing recognition of the technology's potential benefits alongside its risks. This collaboration may set a precedent for future public-private partnerships, fostering a more dynamic regulatory environment that encourages innovation while ensuring safety. However, the trade-offs involved, such as the blocking of benign requests, highlight the complexities of balancing security with usability. As Anthropic and other AI firms navigate these challenges, the industry may see a shift towards more stringent safety protocols, potentially reshaping competitive dynamics and influencing how AI technologies are developed and deployed globally over the next year.
π€ Google DeepMind and A24 announce first-of-its-kind research partnership

- First-of-its-kind collaboration - This partnership between Google DeepMind and A24 is unique in its focus on integrating AI research directly into the creative filmmaking process, setting a precedent for future collaborations in the industry.
- Artist-driven innovation - By involving filmmakers in the development of new technologies, the partnership ensures that innovations are tailored to meet the specific needs and visions of artists, potentially transforming storytelling techniques.
- Investment in creativity - Google's financial investment in A24 underscores its commitment to fostering a creative ecosystem where technology and artistry can co-evolve, enhancing the future of entertainment.
The Bigger Picture: The partnership between Google DeepMind and A24 signals a pivotal shift in how technology and creativity intersect, potentially redefining the landscape of entertainment. By embedding AI research within the creative process, this collaboration not only empowers filmmakers but also positions Google as a key player in the evolution of storytelling. As the entertainment industry increasingly embraces AI, we may see a wave of innovations that enhance narrative depth and audience engagement, while also raising questions about the role of technology in artistic expression. Over the next 6-12 months, this initiative could catalyze similar partnerships across various creative sectors, driving a new era of collaborative innovation.
π’ OpenAI floats giving Trump administration 5 percent cut of AI boom

- 5% government stake proposal - OpenAI's suggestion to provide a 5% stake to the Trump administration indicates a strategic move to align interests and potentially influence regulatory frameworks.
- Regulatory navigation - By engaging with the government, OpenAI aims to preemptively address regulatory challenges that could stifle innovation and growth in the AI sector.
- Industry trend towards partnerships - This move is part of a broader trend where tech companies are increasingly seeking partnerships with governments to secure favorable conditions for their operations.
The Bigger Picture: OpenAI's proposal to offer a stake to the Trump administration underscores a significant shift in how tech companies are approaching regulation in the AI space. By fostering a partnership with government, OpenAI may not only mitigate regulatory risks but also set a precedent for other firms looking to navigate the complex regulatory landscape. This trend could lead to a new era of collaboration between tech giants and governments, where regulatory frameworks are shaped by corporate interests, potentially sidelining public accountability. As the AI boom continues, the implications of such partnerships could redefine the balance of power in the tech industry, favoring those who can effectively engage with policymakers.
ποΈ AI Bytes
π€ Tencent's Apache-licensed Hy3 takes on GLM-5.2 at half the size β and wins everywhere except coding
Tencent has launched Hy3, a 295-billion-parameter Mixture-of-Experts model under the Apache 2.0 license, removing previous regional restrictions and positioning itself as a leader in open-source AI. While Hy3 outperforms GLM-5.1 in various applications, it still trails behind GLM-5.2 in coding tasks, highlighting a strategic trade-off in deployment efficiency and reliability. With significant improvements in hallucination rates and commonsense error rates, Hy3 is tailored for enterprises seeking robust performance without the need for extensive infrastructure.
π€ New Alibaba AI framework skips loading every tool, cutting agent token use 99%
Alibaba researchers have introduced SkillWeaver, a novel AI framework that optimizes tool usage in enterprise workflows by reducing token consumption by over 99%. Utilizing a technique called Skill-Aware Decomposition (SAD), SkillWeaver enhances task accuracy through iterative feedback loops, allowing agents to effectively manage complex multi-step operations. This innovative approach significantly improves the efficiency of AI agents in orchestrating diverse tool ecosystems, addressing a critical bottleneck in enterprise AI applications.
π’ Can Cursor Remain a Platform for OpenAI and Anthropicβs Models Inside SpaceX?
SpaceX's impending acquisition of AI coding startup Cursor for $60 billion raises questions about the platform's future as an open model aggregator for AI labs like OpenAI and Anthropic. While Cursor has historically relied on third-party models, the acquisition could shift its dynamics, potentially limiting access to rival models and intensifying competition in the AI space. The deal also positions Cursor to leverage SpaceX's computing resources, which may enhance its offerings and pricing strategies against major competitors.
π Alibaba reportedly bans employees from using Claude Code
Alibaba has announced a ban on employees using Anthropic's Claude Code, effective July 10, amid concerns over high-risk software classification. This move follows Anthropic's restrictions on Chinese companies accessing its models and efforts to close loopholes that allowed unauthorized access. Employees are now directed to utilize Alibaba's own programming tool, Qoder, instead.
π οΈ Top AI Tools This Week
Descript
Direct your AI co-editor to do your video editing for you, or do it yourself with intuitive editing tools. With Descript, video editing is as easy as typing.
Fireflies
The #1 AI assistant for your meetings. Transcribe, summarize, search, and analyze all your team conversations.

