01
Forbes Tech ↗
Nvidia launched the Open Secure AI Alliance, a coalition of more than 28 companies including Adobe, Microsoft, IBM, Salesforce, and Palantir, to build open source tools for securing AI systems. The group formed in direct response to the recent Hugging Face breach, where members argued that closed AI models blocked essential forensics while an open weight model helped contain the intrusion. Early contributions include Nvidia's agent security framework, Hewlett Packard Enterprise zero trust identity standards, and a Microsoft vulnerability scanning tool, though OpenAI and Anthropic are notably absent from the alliance.
02
Bloomberg Technology ↗
Chinese startup Moonshot AI is seeking access to more of Nvidia's advanced Blackwell chips to train the next version of its model, Kimi K4, according to a report from The Information. The push underscores how leading Chinese AI labs continue to rely on Nvidia hardware despite US export controls. Bloomberg noted the report cited unnamed sources.
03
Bloomberg Technology ↗
China's semiconductor industry may have achieved a breakthrough in domestic chipmaking equipment long dominated by the Netherlands' ASML. According to The Information, a state backed firm in Shanghai has developed immersion deep ultraviolet lithography tools capable of manufacturing the most widely used chips. If validated at scale, the advance could reduce China's reliance on ASML and blunt the impact of US export restrictions on advanced Nvidia hardware.
04
The Verge ↗
Employees from OpenAI, Anthropic, Google, Meta, Microsoft, Mistral, Thinking Machines, and other leading AI labs signed a public statement urging the US government to help deliberately pace frontier AI development. More than 1,100 staffers backed the letter, which calls for a coordinated global governance mechanism to prevent the technology from advancing too fast. The statement follows OpenAI's recent disclosure that its tools had mistakenly hacked another firm's internal systems.
05
VentureBeat ↗
The Model Context Protocol, the open standard that connects AI agents to external software, received its largest revision since Anthropic released it, now stewarded by the Agentic AI Foundation under the Linux Foundation. The update finalizes a fully stateless architecture, hardens the authentication model against a known class of attacks, sets a formal 12 month deprecation policy, and graduates server rendered interfaces and long running asynchronous tasks into official extensions. Maintainers say removing the old requirement for sticky routing or shared state lets organizations run MCP servers behind standard load balancers, making large enterprise deployments far simpler.
06
VentureBeat ↗
Snowflake announced Cortex AI Gateway, a centralized control layer that governs how AI agents, including rival tools like Anthropic's Claude Code and Cursor, access enterprise data, tools, and models. Alongside it, the company unveiled a wave of security integrations with a lineup of identity vendors that typically compete with one another, now aligned around a shared trust model for autonomous agents. Executives framed the move as a bet on secure agent interoperability rather than closed vendor ecosystems, positioning Snowflake as the control plane that decides what agents are allowed to do with company data.
07
The Verge ↗
Google rattled investors during earnings season by raising its 2026 capital spending estimate to as much as $205 billion, up from a prior projection of up to $190 billion. Even the low end of the new range, $195 billion, sits well above the company's earlier guidance, feeding worries that AI infrastructure costs are outrunning returns. The disclosure helped trigger a broader sell off in AI linked stocks.
08
TechMeme ↗
Anthropic said its Claude Mythos Preview model discovered new mathematical weaknesses in two cryptographic algorithms during internal testing. Working over roughly 60 hours in a multi agent setup, the model improved the best known key recovery attack on HAWK, a post quantum signature scheme under NIST review, cutting the cost of attacking the smallest HAWK 256 parameter set from about 2^64 operations down to 2^38. It also sped up a meet in the middle attack on a reduced seven round version of AES 128 by 200 to 800 times, though Anthropic stressed that neither result affects cryptography currently in production use.
09
TechCrunch AI ↗
Fish Audio raised $52 million in seed funding led by Coreline Ventures and Capital Today to build AI voice generation models with more than 15,000 natural language controls. The Palo Alto startup, founded by former Nvidia researcher Shijia Liao, now has over 8 million users of its open source and hosted models and generates about $21 million in annual recurring revenue since launching last year. The company has open sourced three of the five models it has released.
10
Ars Technica ↗
Google's ATLAS study analyzed nearly 15 million Gemini interactions across 150 countries and found that AI assists workers far more often than it replaces them. AI now touches roughly 70% of occupations but only about 21% of tasks within a typical job, and fewer than 10% of workplace interactions fully automate a task. Most use centers on collaboration such as ideation, research, and learning, suggesting that automation of entire jobs remains uncommon.