AI news — March 9, 2026
The 18 top stories PulseAugur surfaced that day, ranked by signal across labs, papers, and developer communities.
-
Anthropic's Claude Code Max compute costs are far lower than reported
A recent analysis disputes claims that Anthropic is losing thousands of dollars per user on its Claude Code Max plan. The author argues that a Forbes report conflated retail API prices with actual compute costs, which are significantly lower. While a heavy user might incur $5,00…
-
Anthropic's Claude Code compute costs are far lower than reported
A recent analysis disputes claims that Anthropic is losing thousands of dollars per user on its Claude Code Max plan. The author argues that a Forbes report conflated retail API prices with actual compute costs, which are significantly lower. While a heavy user might incur $5,00…
-
Anthropic sues US government over AI risk designation; vetting page vanishes
AI firm Anthropic has sued the US government, challenging its designation as a "supply chain risk" after disputes over military use of its AI tools. The company argues the government's actions, including public criticism and contract restrictions, violate its First Amendment rig…
-
Mog programming language enables AI agents to safely modify themselves
Mog is a new programming language designed for AI agents to modify themselves safely and efficiently. It is statically typed and compiled, allowing AI agents to write, compile, and load Mog programs as plugins with controlled function access. The language emphasizes security thr…
-
Anthropic sues Pentagon over AI use restrictions, as Trump admin eyes safety tests
Anthropic is suing to prevent the Pentagon from blacklisting its AI models, arguing the restrictions are unwarranted. Concurrently, the Trump administration is reportedly considering new safety testing requirements for AI models deployed by government agencies. This policy shift…
-
Google DeepMind marks 10 years since AlphaGo's historic Go victory
Google DeepMind is commemorating the 10th anniversary of AlphaGo's historic victory over a world champion Go player. This event marked a significant milestone, demonstrating AI's potential beyond games and paving the way for tackling complex scientific challenges. The techniques…
-
Florida partners with Future of Life Institute to shield families from AI harm
Florida Governor Ron DeSantis has directed state agencies to collaborate with the Future of Life Institute (FLI) on initiatives to protect citizens from AI-related harms. This partnership will result in a training curriculum for crisis counselors on identifying and addressing AI…
-
AI policy and the battle for computing power
Ben Buchanan, an AI policy expert and former White House advisor, discusses the intersection of AI innovation, global power dynamics, and international cooperation. He highlights the strategic importance of computing power and the challenges of AI governance, particularly in the…
-
Nscale secures $790M debt for Norway AI data center, lands Microsoft deal
Nscale, a UK-based AI infrastructure startup, has secured $790 million in debt financing to build an AI data center in Narvik, Norway. This facility was previously intended for OpenAI's Stargate Norway project. Microsoft is set to rent Nvidia chips at this new data center. Nscal…
-
OpenAI acquires Promptfoo to bolster AI agent security and evaluation
OpenAI has announced its intention to acquire Promptfoo, a company specializing in AI security and evaluation tools. This acquisition aims to enhance the security and testing capabilities of OpenAI Frontier, a platform designed for building and deploying AI coworkers. Promptfoo'…
-
AI's labor impact: Jobless growth and cognitive risks emerge
A recent analysis suggests that generative AI is not significantly boosting productivity or creating jobs outside of a few specific roles, despite substantial investments in AI infrastructure. The U.S. labor market is experiencing a "jobless growth" economy, with hiring signific…
-
Two Minute Papers explains physics bug fix using new research
Researchers have resolved a long-standing physics bug that had previously stumped many, detailed in a new paper. This breakthrough addresses a fundamental issue, potentially paving the way for advancements in physics simulations and related fields. The paper, available online, o…
-
Anthropic deemed supply chain risk; 'cancel ChatGPT' trend grows
Anthropic has been designated a supply chain risk by the Department of Defense. Concurrently, a trend of users seeking to 'cancel ChatGPT' is gaining traction, partly fueled by OpenAI's recent agreement with the U.S. military. This newsletter also briefly mentions other AI devel…
-
AI models achieve recursive self-improvement; coding agents shift focus to review
AI models are beginning to autonomously train smaller models, marking a significant step towards recursive self-improvement. This capability, previously confined to major AI labs, is now accessible to individuals with GPUs, allowing for experimentation with models improving mode…
-
mcp2cli offers unified CLI for APIs, reducing token waste
McP2CLI is a new command-line interface tool designed to interact with various APIs, including MCP servers, OpenAPI specifications, and GraphQL endpoints, without requiring code generation. It aims to significantly reduce token usage by abstracting away complex tool schemas. The…
-
Hugging Face introduces Ulysses for training models with million-token contexts
Hugging Face has introduced Ulysses, a novel sequence parallelism technique designed to enable training of large language models with context windows of up to one million tokens. This method addresses the computational challenges associated with processing extremely long sequenc…
-
Hugging Face releases LeRobot v0.5.0 and v3.0 datasets for robotics scaling
Hugging Face has released LeRobot v0.5.0, a significant update to its robotics simulation and data platform. This release focuses on scaling across multiple dimensions, including model size, dataset volume, and simulation complexity. Alongside the platform update, Hugging Face a…
-
Languages are dying out, but new ones are being born through technology and subcultures
While hundreds of languages have become extinct since 1960, a process accelerated by globalization and technology, new linguistic diversity is also emerging. The same forces that break down traditional language barriers also foster the creation of specialized vocabularies within…