hodler

30 Jan 2026, 12:54

Anthropic disputes Pentagon over military AI scope in $200M contract

The Pentagon and Anthropic are in a direct fight over how artificial intelligence can be used by the U.S. military. The conflict centers on safeguards that would block the government from using AI to target weapons on its own or to run surveillance inside the United States. The disagreement sits inside a contract valued at up to $200 million and has now stalled talks. This dispute has become an early test of how much influence Silicon Valley really has in Washington after years of tension. Defense and intelligence officials want freedom to deploy stronger AI tools in combat and security work. Tech leaders want limits. The talks have dragged on for months and have now hit a wall. Pentagon presses ahead as Anthropic pushes back on weapons use After long negotiations, the U.S. Department of Defense and Anthropic are stuck. Six people briefed on the talks said neither side has moved. The clash has grown sharper under President Donald Trump’s second term, with disagreements inside the administration now spilling into public view. In a statement, Anthropic said its technology is “extensively used for national security missions by the U.S. government and we are in productive discussions with the Department of War about ways to continue that work.” At the same time, company representatives told officials they worry the tools could be used to spy on Americans or help weapons strike targets without enough human control. Pentagon leaders rejected those limits. They pointed to a January 9 memo on AI strategy that says the military should be free to use commercial AI systems as long as the law is followed. Officials said private rules should not decide battlefield choices. Even so, the Pentagon still needs Anthropic to move forward. The models are built to avoid actions that could cause harm. Company engineers would have to adjust the systems before the military could use them the way it wants. The standoff puts Anthropic’s defense business at risk during a sensitive moment. The San Francisco startup is preparing for a future public offering. It has spent heavily to win U.S. national security work and to shape federal AI policy from the inside. Anthropic is also one of only a few firms the Pentagon selected last year. Others include Google, Elon Musk’s xAI, and OpenAI . These companies now sit at the center of U.S. military AI plans. Caution from Anthropic has caused friction with the Trump administration before. In a blog post this week, CEO Dario Amodei warned that AI should support national defense “in all ways except those which would make us more like our autocratic adversaries.” Dario has also spoken out on government force at home. After fatal shootings of U.S. citizens during immigration protests in Minneapolis, he described the deaths as a “horror” in a post on X. The smartest crypto minds already read our newsletter. Want in? Join them .

30 Jan 2026, 12:42

Circle maps out 2026 strategy to grow USDC usage

Circle Internet Group says it will devote 2026 to strengthening its underlying infrastructure to enable more businesses and institutions to adopt stablecoins. Nikhil Chandhok, the firm’s chief product and technology officer, noted in a blog post that the company is trying to bring Arc, its institutional-focused layer-1 blockchain, out of testnet and into the real world. He noted the company will also focus heavily on upgrading stablecoin infrastructure so that companies can adopt stablecoin-based payments and settlements without building their own systems from scratch. The company also said it will deepen the utility of USDC , EURC, USYC, and partner tokens while extending their presence to new chains. Arc will serve as Circle’s coordination layer Circle’s decision follows strong growth in stablecoins, now worth more than $300 billion , along with clearer U.S. regulatory frameworks. Stablecoins became a major focus of the crypto industry in 2025, driven by new U.S. rules and growing interest from banks and institutions. The plan, Circle says , is to scale applications , including its payments network, so organizations can handle stablecoin payments without managing the technology themselves. Chandhok even noted that Circle will invest in making USDC more seamless across different blockchains, while improving the user and developer experience. He further stated, “In addition, we will continue to expand our partner and developer ecosystem to build utility and extend global scale and reach to bring the benefits of stablecoin and internet-scale finance to more markets and use cases.” In the first 90 days of operation, the Arc testnet generated almost 1.5 million wallets that processed more than 150 million transactions, typically settling the transactions in about half a second. Circle hopes to have Circle Payments Network and StableFX run on Arc Services like Circle Payments Network and StableFX will in the future operate on Arc, drawing on its institutional-grade architecture to more easily coordinate payments and capital flows, Circle said. Circle also seeks to have Arc lower onchain finance’s barriers to entry by handling behind-the-scenes complexity for regulated organizations. The platform also comes with developer and interoperability tools to help teams build leading product solutions. For instance, the company’s CCTP has also supported cross-chain USDC, with the stablecoin running across 30 networks by December 2025 and CCTP connecting 19 of them, processing $126 billion in total. Speaking on CCTP, the company even stated in its release that it’s “Going forward, our priority is to make it an even more systemic interoperability layer for USDC so that businesses and users have access to USDC liquidity that moves safely and predictably across blockchains where it’s needed.” The firm also recently launched Circle Gateway for chain-agnostic USDC balances, enabling instant cross-chain liquidity for apps without increasing complexity. Currently, it’s infusing Gateway with Arc, CCTP, and x402, preparing the base for USDC-powered micropayments, machine-to-machine transactions, and agentic payment flows across multiple chains. The firm wrote, “Together with Arc, these interoperability and developer tools will make the idea of an Economic OS more than just a concept, but rather a robust framework that combines a network, interoperability primitives, and tightly integrated tools for developers to build with.” DeFiLlama data indicates that USDC still holds the No. 2 spot among dollar-backed stablecoins, with a market cap above $70 billion. Meanwhile, USDT remains the largest stablecoin, accounting for more than $186 billion of the $306 billion total market cap. Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free .

30 Jan 2026, 12:20

Circle Expands Stablecoin Reach, Targets Institutional Adoption

Circle Internet Group plans to spend 2026 improving its technology base to support stablecoin use among businesses and institutions.

30 Jan 2026, 12:05

Former Google engineer convicted of stealing 2,000 AI trade secrets for China

A US federal jury found a former Google engineer guilty of stealing artificial intelligence trade secrets and sending them to China, according to the Department of Justice’s statement released Thursday. The 38-year-old defendant, Linwei Ding, has been convicted of seven counts of economic espionage and theft of trade secrets. The US Northern District of California court ruling said that the actions were carried out for the benefit of the People’s Republic of China. During the trial, Ding was accused of stealing more than 2,000 internal documents from Google’s systems. US Attorneys, in conjunction with the Federal Bureau of Investigation found that the files were transferred to his personal Google Cloud account while he was still on the company’s payroll. The Justice Department first brought the charges in March 2024, but a later indictment added more counts to the allegations, including claims that Ding participated in Chinese AI technology initiatives. Former Google engineer sent data center schematics to Beijing According to the evidence presented in court, the stolen documents contained plans for Google’s advanced computing infrastructure. The material included data center schematics capable of providing sufficient power for large artificial intelligence projects. The stolen information also provided information about Google’s internal software for managing computing clusters. That software coordinates thousands of specialized chips into a unified system, which is purportedly central to the company’s AI capabilities. Jurors heard that the files contained technical details on proprietary hardware, including Google’s Tensor Processing Units and Graphics Processing Units. The data also covered how the software used in those chips communicates and executes several complex tasks. Another leaked topic was Google’s SmartNIC technology, a specialized network interface card that supports communication within AI supercomputers , cloud networks, and other services. According to witnesses’ testimonies, the document transfers happened between May 2022 and April 2023. Ding was an employee at Google at the time, while he was also building connections with companies based in China. Prosecutors said Ding was in talks to become a chief technology officer at a PRC technology startup. By early 2023, he was working to establish his own AI and machine learning company in China as the company’s chief executive. In presentations to investors, Ding allegedly said he could replicate advanced AI computing systems by adapting Google’s technology. Ding downloaded the material onto his personal computer less than two weeks before his resignation in December 2023, per court records . Evidence citing Ding’s interactions with the Chinese government showed he applied to a Shanghai-based government-backed talent program in late 2023. “Ding’s application for this talent plan stated that he planned to help China have computing power infrastructure capabilities that are on par with the international level. The evidence at trial also showed that Ding intended to benefit two entities controlled by the government of China by assisting with the development of an AI supercomputer and collaborating on the research and development of custom machine learning chips,” the DOJ wrote in its statement . National security concerns emerge as the AI race continues US officials said Linwei Ding’s actions and the misuse of AI research pose risks to America’s national security. According to the FBI and DOJ, Silicon Valley is pioneering AI research that would boost the country’s economic growth and improve its security. “The theft and misuse of advanced artificial intelligence technology for the benefit of the People’s Republic of China threatens our technological edge and economic competitiveness,” said FBI Special Agent in Charge Sanjay Virmani. Ding is scheduled to appear at a status conference on February 3, where he will be sentenced. He is facing a potential 10-year sentence for each count of trade secret theft, while each economic espionage conviction could bring up to 15 years in prison. Meanwhile, China has been investing heavily in AI infrastructure since 2021, directing around $100 billion into AI data centers. However, a recent industry report said the average utilization rate nationwide is just 32%. In an opinion article published in China Economic Weekly, Rao Shaoyang of the China Telecom Research Institute warned the country against “blindly building intelligent computing centres” and asked planners to look at local demand before launching any new projects. The smartest crypto minds already read our newsletter. Want in? Join them .

30 Jan 2026, 10:39

Microsoft unveils touch-sensing system to overcome key robot limitations

Microsoft Researc h ro lled out a new robot control system in late January 2026 that lets machine s wo rk with their hands while processing spoken commands and physical feedback. The system, called Rho-alpha, marks the company’s entry into foundation models designed for robots that use two arms at once. The technology will first reach select groups through an Early Access Program before Microsoft makes it available more widely on its Foundry platform. Companies can then adapt the system to their specific needs using their own data. Adding touch to robot intelligence Factories and warehouses are looking for robots that can handle changing conditions rather than repeating the same programmed motions forever. Hospital settings need machines tha t ad just to different situations. Production lines where items vary from batch to batch create problems that old-style automation can’t solve efficiently. Microsoft built Rho-alpha to fill this need by processing what robots see and hear alongside what they physically feel through sensors. Most robot systems today rely on cameras and microphones to understand their surroundings and take instruction. Rho-alpha adds another layer by treating touch as equally important. When a robot gripper has pressure sensors built in, the system gets information that cameras miss entirely. This matters when trying to plug something into a socket or fit parts together where sight alone doesn’t provide enough detail about whether things are lining up correctly. Microsoft showed off these abilities using two Universal Robots UR5e arms equipped with sensors that detect pressure and contact. During tests with a task set called BusyBox, people told the robot to do things like put a tray inside a toolbox and shut the lid. The system turned those words into coordinated movements between both arms and made adjustments based on what the sensors felt. When attempts to insert a plug didn’t work on the first try, a human operator could guide the robot using a 3D input device, and the system learned from those corrections. Getting enough training data remains the biggest challenge in building capable robots. Language models can learn from massive amounts of text available online, but robot training requires actual physical demonstrations that take time and money to record. Microsoft addressed this by training Rho-alpha on three types of information: recordings of real physical demonstrations, simulated practice tasks, and large datasets of images with questions and answers from the web. The company uses Nvidia Isaac Sim running on Azure servers to create realistic synthetic scenarios through a reinforcement learning process. This simulation setup produces physically accurate practice situations that supplement the real demonstrations. The combined approach lets the mode l en counter unusual cases and failure situations that would otherwise require thousands of hours of real-world operation to capture. The training metho d fo llows pattern s ot her companies i n ro botics are using. Google DeepMind’s Gemini Robotics system, Figure AI’s Helix model for humanoid robots, and Physical Intelligence’s Pi-zero all take similar approaches to work around the data shortage problem. The technique helps these system s le arn general manipulation skills without needing specific demonstrations for every single task they might face. Competing in a maturing market Microsoft joins a ro botics foundation model market that has grown considerably over the past year and a half. Nvidia released GR00T N1.6 aimed at humanoid robots, focusing on whole-body control and understanding context. Google DeepMind expanded Gemini into robotics with abilities ranging from folding paper into origami shapes to handling playing cards. Physical Intelligence presents Pi-zero as an all-purpose system trained across different robot types. Rho-alpha stands out in thre e wa ys. First, the emphasis on tactile sensing tackles situations where systems relying only on vision struggle. Second, the model comes from Microsoft’s Phi series, which the company has tuned to run efficiently on regular consumer hardware. This background suggests it could run on local devices without needing constant connection to cloud servers. Third, the focus on learning from human corrections during actual operation sets it apart from models that need complete retraining to pick up new behaviors. Microsoft’s business approach also differs fro m co mpetitors. The company plans to offer Rho-alpha through its Foundry platform as infrastructure that manufacturers and system integrators can customize with their own proprietary information. This mirrors the company’s approach with Azure OpenAI Service and targets organizations wanting to create specialized versions rather than using a generic model. For manufacturers and logistics companies, the immediate chance lies in spotting repetitive handling tasks where current automation comes up short. Quality inspection stations, operations that assemble kits of items, and small-batch assembly lines represent situations where Rho-alpha’s mix of language understanding and touch sensing could cut down on programming requirements. The early access program Microsoft announced gives organization s a way to test whether the system fits their needs before investing in deployment infrastructure. Companies should enter these evaluations expecting that human supervision will be necessary and should plan for workflows where operators correct and guide the robots through initial learning periods. Physical AI represents a shift from robots as programmed tools to robots as flexible collaborators. That shift will take years rather than months, but the foundation models coming from Microsoft, Nvidia, and Google establish the basic patterns that will define enterprise robotics for the next ten years. Join a premium crypto trading community free for 30 days - normally $100/mo.

30 Jan 2026, 10:30

Never Sell Your Bitcoin: Sats Terminal Founders on Securing Coinbase & Binance Backing, Bitcoin Loans and More

Sats Terminal is the first native Bitcoin super app, bringing together Bitcoin loans, yield, and trading in a single interface and developer SDK. Sats Terminal is backed by YZi Labs (formerly Binance Labs), Coinbase Ventures, and Draper Associates. The founders of Sats Terminal recently joined the Bitcoin.com News Podcast to talk about the technology: Stan