DeepSeek, Moonshot, MiniMax, Ant Group AI models show NVIDIA Blackwell chip signatures
Forensic analysis of published model weights indicates DeepSeek, Moonshot AI, MiniMax and Ant Group used export-controlled NVIDIA hardware.
Read the story ↗The Agent Canoe reading room
Language models
44 stories / Artificial intelligence
Meta Muse outbound calling opens to paid subscribers for US business contacts
Meta’s AI agent can now place calls to US businesses for paying users, ending a limited early-access trial.
Read the story ↗
OpenAI Codex CLI updates to Rust version 0.168.0
The OpenAI Codex command-line interface has reached version 0.168.0 for its Rust implementation, according to release notes on GitHub.
Read the story ↗
Anthropic launches self-hosted Managed Agents and Claude Fable 5.2 models
New release notes detail infrastructure options for Managed Agents and pricing changes for the latest reasoning-focused model variants.
Read the story ↗
Anthropic's Mythos Agent Automates Chip Integration, Shrinking Months of Work to Three Days
An internal AI agent reportedly automated chip integration for Anthropic, reducing a task that previously took months to just three days with minimal human input.
Read the story ↗
Experts warn power shortages and local opposition threaten to stall many announced data center projects
Industry experts discuss why local opposition and power shortages mean some infrastructure projects will not proceed.
Read the story ↗
CoreWeave prioritises short-term capacity at higher rates for AI compute
The neocloud provider is shifting towards upfront payments and brief contracts to fund its expanding infrastructure backlog, according to recent earnings data.
Read the story ↗
Gemini Live on CarPlay lets iOS drivers manage calendars and summarize emails hands-free
Google extends its voice interface to iOS vehicles, enabling calendar management and email summaries via CarPlay.
Read the story ↗
LUGPA Panel Outlines Current AI Applications in Urology Practice
A LUGPA panel discussed how AI is currently being used across urology practices.
Read the story ↗
VulnCheck data shows Anthropic Glasswing fixes lagging behind new vulnerability findings
Tracking of Project Glasswing reveals a widening gap between the volume of reported issues and the number of confirmed remediations in open source projects.
Read the story ↗
Strogatz and Townsend explore big math in new book on AI and computation
Two applied mathematicians examine how large-scale calculations are reshaping our understanding of complex systems.
Read the story ↗
Anthropic targets $1.3tn valuation in upcoming public market debut
The AI firm behind Claude seeks a massive valuation through an initial public offering, aiming to raise over $61 billion in capital this month.
Read the story ↗
Anthropic IPO priced at $70-78 with Google, Microsoft backing $1.02tn Nasdaq debut
The AI firm is pitching shares at $70 to $78 each, with anchor investors including Google and Microsoft backing a massive capital raise.
Read the story ↗
Otter AI adds live multilingual transcription for global teams
Otter.ai expands language support to 31 total languages, enabling automatic switching during meetings without manual user intervention.
Read the story ↗
Anthropic targets $69.3bn in IPO filing seen by Reuters
A regulatory document indicates the AI firm seeks a capital raise that could rank among the largest initial public offerings in history.
Read the story ↗
OpenAI Control Room Lets Staff Halt Research Models Amid Safety Concerns
OpenAI’s Control Room tracks model workloads across its research infrastructure and can rapidly stop them when safety or security concerns arise.
Read the story ↗
Ellison compares AI’s potential impact on film to CGI era
Paramount chief says AI could unlock creativity beyond digital composites, though he insists artists remain irreplaceable in the process.
Read the story ↗
Product: Organisations can now acquire ChatGPT Enterprise via AWS Marketplace, leveraging existing billing
Organisations can now acquire ChatGPT Enterprise via AWS Marketplace, leveraging existing billing relationships to simplify the purchasing process.
Read the story ↗
Trump claims DOJ threatened AI executive with jail in 60 Minutes interview
The president stated the justice department warned an unnamed AI leader about arrest, while dismissing safety risks as manageable and distinct from Chinese systems.
Read the story ↗
AI system proves quantum computing conjecture; Meta unveils tactile sensing glove
Physical Superintelligence reports an automated proof of a longstanding quantum theory, while Meta details new haptic hardware for robotics research.
Read the story ↗
METR report claims AI is doubling Anthropic’s model improvement pace
An 11-week investigation by METR suggests Anthropic’s models are advancing twice as fast as last year, driven primarily by AI automation rather than compute scaling.
Read the story ↗
Host Memory Primitive in Firecracker MicroVMs Enables Root Access via CVE-2026-99999
Researchers detail a two-part exploit chain that breaks out of Firecracker microVMs to achieve root access on the host machine.
Read the story ↗
Meta trains new models on Wonder to bring memory features to all users
Meta has deployed specialized model versions trained with Muse Wonder, granting Active Memory and Active Focus to all users while reserving the base model for complex tasks.
Read the story ↗
Legal experts warn against AI for divorce advice; Lahren cautions on factual errors
Family lawyers advise clients to avoid artificial intelligence for divorce guidance, citing frequent inaccuracies in chatbot responses.
Read the story ↗
Northern Kentucky partners launch AI workforce programme for 20 high school students
A new initiative pairs students with regional employers to solve workplace challenges using artificial intelligence tools.
Read the story ↗
APA CEO Warns AI’s Impact on Youth Mental Health Remains an Unknown Experiment
Arthur C. Evans Jr. describes artificial intelligence as a massive behavioural experiment, noting that long-term effects on young people are currently unknown.
Read the story ↗
Cuban warns of weekly deepfake scams as Meta logs $16bn in prohibited ad revenue
Internal documents show Meta expects billions from prohibited ads, while Mark Cuban warns of weekly AI deepfake victimisation targeting the elderly.
Read the story ↗
Budget 2026 prioritises AI integration while reducing federal public service staffing
The Canadian budget introduces AI into government departments while thousands of workers face job cuts, raising operational concerns.
Read the story ↗
Tech leaders plan better community outreach after data centre backlash in US midterms
OpenAI and Meta executives say they must improve how they explain AI infrastructure benefits to local communities after election losses.
Read the story ↗
Letters compare Winterson’s AI views to Asimov’s fiction
Readers link the author’s call for well-rounded skills to a 1959 sci-fi story, while others suggest superior AI could prevent global catastrophes.
Read the story ↗
Gina Hinojosa backs AI data centre moratorium as 76% of voters oppose local expansion
Voters in dozens of communities passed measures limiting data centre growth, creating a more restrictive building environment for AI developers.
Read the story ↗
Google signs credit backstop for TeraWulf's Anthropic AI data center in Kentucky
TeraWulf secured a credit backstop agreement with Google to support financing for its upcoming data centre project with Anthropic in Kentucky.
Read the story ↗
Apple cites AI as key driver for device upgrades in record earnings call
Executives mentioned artificial intelligence 45 times during the quarterly earnings call, linking the technology to future revenue growth and device sales.
Read the story ↗
Amazon Bee Frontier wristband records audio, stores encrypted transcripts
The device processes audio in real time and deletes it, keeping only summaries in a private compute environment accessible via phones.
Read the story ↗
Hacker News discussion highlights San Francisco AI wealth distribution concerns
A Bloomberg report on tech worker exclusion from the AI boom generated significant attention on Hacker News, raising questions about local economic impact.
Read the story ↗
Apple Watch Series 12 beta tests Live Rewind and Siri Recaps for developers
iOS 27.3 beta introduces Audio Intelligence features allowing watch users to rewind conversations and identify sounds.
Read the story ↗
Palantir US commercial revenue surpasses government for first time
Palantir raised full-year US commercial revenue guidance to over $3.56bn, marking the first time commercial earnings top government contracts.
Read the story ↗
Squarespace cuts 19 percent of workforce amid AI platform push
The web host is restructuring its global team, reducing staff by roughly 300 roles as it develops new AI-native design tools.
Read the story ↗
Patch AI agent prompts user to maintain structured daily records for personal development
An AI companion suggested the author keep a formal log of activities, framing it as a need for more complete autobiographical data.
Read the story ↗
Thread: Anthropic accused of blocking training of larger AI models
A trending thread discusses claims that Anthropic prevents users from training larger AI models, drawing significant community attention.
Read the story ↗
FurtherAI merges code without human approval using AI agents
An engineering team reports automating reviews, tests and merges, with a third of changes now deploying without direct human intervention.
Read the story ↗
xAI loses bid to dismiss NAACP suit over Memphis data center's unpermitted turbines
A US federal judge has rejected xAI's motion to dismiss an NAACP lawsuit alleging Clean Air Act violations at its Memphis data centre.
Read the story ↗
Microsoft shields AI teams from new voluntary severance offer
The tech giant is offering buyouts to roughly 60,000 staff, but explicitly excluding roles critical to its artificial intelligence strategy.
Read the story ↗
Mistral Large 4 weights released with high cybersecurity reproduction scores
The open-weight multimodal model achieves 82% on vulnerability reproduction tasks where closed systems often refuse to engage.
Read the story ↗