News · earlier edition
Over the last 24 hours, OpenAI and Anthropic dominated the news with funding and IPO moves, while new research on AI-driven AI development, cyber capabilities, and chip export controls drew attention.
38 stories ranked · updated Nov 7, 23:03 UTC
OpenAI has secured $11 billion in fresh capital, the first tranche of a round valuing the company at $1.25 trillion. The full round could grow to $20 billion.
Anthropic is moving forward with its IPO despite the tech selloff, at a valuation of $1.25 to $1.75 trillion, below the hoped-for $2 trillion. It aims to raise $80 to $100 billion before Thanksgiving.
Exponential Time Hypothesis co-proposer Russell Impagliazzo called OpenAI's disproof shocking news. The result could eventually speed up chip design and affect encryption, according to the New York Times.
A new report describes Anthropic using AI to build more capable AI, with its models improving roughly twice as quickly as a year ago. The Verge reports researchers are alarmed by the trend.
In a Vals AI Benchmark: A standard test used to compare models, like an exam. Useful, but labs tend to show the tests they do well on, so one score never tells the whole story., OpenAI's GPT-6 Sol Daybreak hit 80.8 percent of attack objectives in simulated corporate networks. The Open weights: The model's files are public, so anyone can download it, run it on their own computer and build on it, usually for free. trckr labels these "Open". DeepSeek V4.1 Flash reached 72.8 percent at a tenth of the cost.
A New York Times investigation found some Chinese firms approved by the Commerce Department to handle Nvidia H200 chips have empty offices and ties to sanctioned firms. The Verge reports concerns about the vetting process.
Google Research built an AI that spots signs of high blood pressure from the pulse and motion sensors in a standard smartwatch. In a 196-person study, it detected 65.8 percent of cases, beating the traditional approach according to The Decoder.
Google has acquired Crusoe's data center site in southeast Wyoming. Crusoe reportedly failed to lock in customers, including Google, for the project.
Meta introduced specialized versions of its existing models, trained using Muse Wonder, and is bringing Active Memory and Active Focus to all Muse and Meta AI users, per The New Stack.
Security firm Apiiro found secrets such as passwords leaked into 1,122 public GitHub files through Claude Code's Don't Ask Again option. The report was widely discussed on Hacker News.
Releases
NVIDIASpecialist
MinorMinor update: A small change to an existing model: a dated version, a preview refresh or a variant.OpenOpen model: Its files are public (open weights), so anyone can download it and run it on their own computer, usually for free.NVIDIA's robot-training example, free to download, that teaches a simulated robot to lift, hold and put back a tissue box. Made for robotics researchers.
TencentText & chat
MinorMinor update: A small change to an existing model: a dated version, a preview refresh or a variant.OpenOpen model: Its files are public (open weights), so anyone can download it and run it on their own computer, usually for free.Tencent released a free-to-download helper model that speeds up Alibaba's Qwen3.8-27B by drafting text in parallel, mainly for developers running AI at scale.
Turn on the daily digest for any channel: email, Telegram, Discord, push or ntfy.
Archive
NVIDIA
Google
Meta