The lunch briefing.
Mid-day check: Anthropic makes a strategic acquisition, DeepSeek slashes AI model prices, and the FBI seeks real-time access to license plate readers.
AI Market Competition & Investment. DeepSeek's aggressive 75% price cut on its flagship V4 model signals intensifying competition in the AI model market, potentially pressuring rivals to adjust pricing. Anthropic's $300 million acquisition of Stainless, a startup specializing in SDK generation, highlights a trend of major AI players consolidating talent and technology. This strategic move aims to enhance Anthropic's developer tools and accelerate its competitive edge against industry giants. The acquisition underscores the high stakes and rapid pace of investment in the frontier AI space.
AI Policy & Societal Impact. The US government's approach to AI regulation faces scrutiny after President Trump reversed a planned safety review of new AI models, influenced by big tech lobbying. This decision raises concerns about unchecked AI power and its potential risks. Separately, UC Berkeley Law is implementing a strict ban on AI use in graded work, emphasizing foundational human reasoning skills. These developments reflect ongoing debates about responsible AI development and its ethical implications.
Core Infrastructure & Hardware. The escalating demand for AI is driving significant investment in underlying infrastructure, with Europe, Meta, and Japan racing to deploy the first petabit-class submarine cable by 2030. This global effort supports the massive data transfer requirements of advanced AI systems. Concurrently, Alibaba's Qwen model demonstrated autonomous code optimization for its custom chip over 35 hours, showcasing advancements in AI-driven hardware development. Samsung's chairman also met with MediaTek to discuss strategic memory chip and foundry orders.
AI in Practice & Security. AI's practical applications continue to expand, from enhancing fan experiences with IBM's AI at Ferrari to the rising demand for forward-deployed engineers for AI agent integration. However, these advancements also bring new security challenges, as evidenced by the "Megaladon" attack compromising over 5,500 GitHub repositories. Concerns about privacy and surveillance are highlighted by the FBI's push for near real-time access to US license plate readers. These incidents underscore AI's dual nature as both an enabler and a potential vulnerability.
Anthropic's $300M Stainless deal lands hardest on OpenAI and Google
Anthropic acquired Stainless, a New York startup whose software generated SDKs for major AI companies. This $300 million deal is expected to impact OpenAI and Google significantly by strengthening Anthropic's ecosystem.
DeepSeek permanently reduces price of its flagship V4 model by 75 percent
DeepSeek has permanently cut the price of its flagship V4 AI model by 75 percent, a move that could be aimed at undercutting competitors in the market. This significant price reduction makes the model more accessible to developers and businesses.
How big tech got its way on Trump’s AI executive order
President Trump abruptly backed out of signing an executive order that would have called for a government safety review of new AI models. This reversal followed lobbying efforts by big tech companies.
Alibaba's latest AI model ran autonomously for 35 hours to optimize code for its own custom chip
Alibaba's Qwen team released Qwen3.7-Max, a proprietary model built for long-running autonomous agent tasks, which optimized code for its custom chip for 35 hours. The model matches Claude Opus 4.6 on benchmarks and beats Chinese rivals.
Europe races Meta and Japan to launch first Petabit-class submarine cable before 2030
IOEMA-1 is advancing Northern European subsea cable development, while Japan and Meta explore high-capacity systems for global internet expansion. This race is driven by exploding AI demand for data transfer.
New attack "Megaladon" compromises 5.5K+ GitHub repos
A new attack dubbed "Megaladon" has compromised over 5,500 GitHub repositories. This incident highlights ongoing security vulnerabilities in code hosting platforms.
UC Berkeley Law draws a hard line against AI in legal education
UC Berkeley Law will ban AI from nearly all graded work starting in summer 2026, including outlining, drafting, and proofreading. The school believes future lawyers must first learn to think for themselves before using AI meaningfully.
People used AI to recreate the voices of pilots killed in a plane crash
Individuals used AI to recreate the last few seconds of a conversation between pilots killed in a plane crash, based on a PDF report. This raises ethical questions about the use of AI for sensitive historical recreations.
Google CEO Pichai now calls links a "part" of search
Google CEO Sundar Pichai now refers to links and sources as merely a "part" of search, signaling a shift from traffic distributor to an AI publisher. This redefinition suggests new features will keep users within the Google ecosystem.
Microsoft’s new responsible tech lead on humanizing high-speed AI development
Jenny Lay-Flurrie, head of Microsoft's Trusted Technology Group, emphasizes building AI responsibly and maintaining that standard. Her role focuses on humanizing the rapid development of AI technologies.
The FBI wants 'near real-time' access to US license plate readers
The FBI is seeking near real-time access to US license plate readers, raising concerns about privacy and surveillance. This initiative aims to enhance law enforcement capabilities through widespread data access.
Plastic surgeons asked to create ‘AI face’ as clients seek unrealistic looks
Plastic surgeons are increasingly concerned about clients seeking "AI face," presenting unrealistic AI-generated visions of desired appearances. This trend highlights the growing influence of AI on societal beauty standards and expectations.
Forward Deployed Engineers are becoming AI’s hottest jobs
As enterprises prepare to deploy AI agents across various workflows, companies are realizing the critical need for forward-deployed engineers. These roles are becoming highly sought after for integrating AI solutions effectively.
Samsung chairman secretly visits MediaTek seeking to trade memory chips for foundry orders
Samsung Electronics Chairman Lee Jae-yong reportedly visited MediaTek to discuss trading memory chips for foundry orders. This discreet meeting signifies strategic negotiations in the global semiconductor supply chain.