The morning briefing.
While you slept: OpenAI acquires a voice cloning startup, Cerebras eyes a massive IPO, and one in seven Brits are using ChatGPT for medical advice.
AI Market Dynamics. OpenAI's acquisition of Weights.gg, a voice cloning startup, signals continued consolidation in the AI sector, focusing on specialized capabilities. Meanwhile, Cerebras is reportedly eyeing a massive $60 billion IPO, highlighting the significant capital flowing into AI hardware. This financial activity underscores the intense competition and high stakes in the frontier AI market.
Agentic AI Costs and Capabilities. The OpenClaw founder's reported $1.3 million monthly spend on 100 AI agents demonstrates the escalating operational costs for advanced AI development. Despite these investments, a new benchmark reveals that AI video generators still struggle with logical reasoning, indicating a gap between stunning visual output and true world understanding. This suggests that raw compute power alone isn't solving fundamental AI challenges.
AI in Society and Policy. The increasing integration of AI into daily life is evident as one in seven Brits use ChatGPT for medical advice, raising questions about public trust and regulatory oversight. YouTube's expansion of its AI deepfake detection tool to all creators reflects growing concerns over synthetic media. These developments highlight the urgent need for robust policy frameworks and ethical guidelines for AI deployment.
AI Research and Development. New research from the Allen Institute for AI shows a mixture-of-experts model achieving near-full performance with only 12.5% of its experts, potentially making MoE models more practical. Separately, Google is pushing back on "generative engine optimization" and "answer engine optimization," asserting that traditional SEO principles still apply to AI search. These efforts aim to refine AI efficiency and clarify industry best practices.
Cerebras' $60B IPO: Slowly, then All at Once
Cerebras Systems, a company specializing in AI supercomputing chips, is reportedly preparing for an initial public offering that could value it at $60 billion. This potential IPO highlights the significant investment and growth in the specialized AI hardware sector.
New benchmark confirms AI video generators look stunning but still can't reason about the world
A new benchmark, WorldReasonBench, evaluates AI video generators on physical and logical plausibility, not just image quality. ByteDance's Seedance 2.0 leads, but all models, including commercial ones, struggle significantly with logical reasoning tasks.
OpenAI bought a voice cloning startup famous for celebrity imitations
OpenAI has acquired Weights.gg, a small startup known for enabling users to create AI voice clones of celebrities. The team has joined OpenAI, but the company has no immediate plans to release a standalone voice cloning product.
Forward deployed engineer is AI’s hottest job as OpenAI and Google race to hire
The role of a forward deployed engineer has become one of the most sought-after positions in AI, with companies like OpenAI and Google actively recruiting for these roles. This trend reflects the growing need for engineers who can bridge the gap between AI development and real-world application.
For $1.3 million a month, OpenClaw founder runs 100 AI agents that code, review PRs, and find bugs
Peter Steinberger, founder of OpenClaw, is spending $1.3 million monthly on OpenAI API calls to run 100 Codex instances for his open-source project. He views this significant expenditure as a research investment into the future of software development where token costs are not a limiting factor.
People are building chatbot versions of their ex-partners
Individuals are reportedly creating AI chatbot replicas of their ex-partners, raising ethical questions about the use of AI in personal relationships and emotional healing. A psychologist expressed skepticism about the therapeutic benefits of such practices.
One in seven Brits swapped their GP for ChatGPT, study finds
A new study indicates that one in seven people in Britain have used ChatGPT for medical advice, bypassing traditional general practitioners. This trend emerges as the NHS continues to deliberate the appropriate integration of AI into healthcare.
Researchers train AI model that hits near-full performance with just 12.5 percent of its experts
Researchers at the Allen Institute for AI and UC Berkeley developed EMO, a mixture-of-experts model that retains nearly full performance even when 75% of its experts are removed. This breakthrough could make MoE models more practical for memory-constrained environments.
YouTube's AI deepfake detection tool is now available to all creators 18 and older
YouTube has expanded access to its AI deepfake detection tool, making it available to all creators aged 18 and older. This move aims to help content creators identify and manage synthetic media on the platform.
Frontier AI has broken the open CTF format
Frontier AI capabilities are reportedly disrupting the traditional open Capture The Flag (CTF) format, challenging the established methods for evaluating and securing systems. This suggests a need for new approaches in cybersecurity challenges as AI advances.
Google says GEO and AEO are a myth and traditional SEO is all you need for AI search
Google has stated that new SEO buzzwords like "generative engine optimization" (GEO) and "answer engine optimization" (AEO) are simply rebranded traditional SEO. The company's new documentation clarifies that AI search relies on the same ranking systems as conventional search.
Brussels runs on summaries and AI is now writing them
The European Union is increasingly utilizing artificial intelligence to generate summaries, addressing its long-standing challenge of information overload. This adoption signifies a shift towards AI-powered solutions for administrative tasks within the EU.
Europe built sovereign clouds to escape US control. Then forgot about the processors
Europe's efforts to establish sovereign clouds to reduce reliance on US control have overlooked a critical component: the underlying processors. The lack of independent certification for silicon layers like Intel ME and AMD PSP poses a security and sovereignty challenge.
Microsoft lets Insiders choose their Windows 11 Start menu size
Microsoft is rolling out new customization options for Windows 11 Insiders, allowing users to choose the size of their Start menu. The company is also testing the ability to reposition the Windows 11 taskbar, offering more flexibility for user interface preferences.