NerdNewsApril 13, 2026 |
|
News & Trends
OpenAI Says Elon Musk Is ‘Injecting Chaos’ with Recent Legal Maneuver
OpenAI has accused Elon Musk of 'injecting chaos' into their ongoing lawsuit by recently amending his claim to seek $134 billion in damages, which would be paid to OpenAI's nonprofit instead of himself. This move has been characterized as a 'legal ambush' by OpenAI, occurring just weeks before the trial is set to start.
Microsoft Says You Should Take Copilot Seriously but Not Literally
Microsoft has stated that users should take its AI assistant Copilot seriously, but not literally, as it evolves beyond its initial 'entertainment purposes only' label. The company plans to update its terms of use to reflect Copilot's growing capabilities and uses, acknowledging its potential impact on daily workflows and decision-making.
At the HumanX conference, everyone was talking about Claude
At the HumanX AI conference, Claude was the most talked about chatbot, surpassing ChatGPT. OpenAI's recent controversies and vacillations have led to a perception that the company is losing its footing, while Anthropic's Claude is gaining traction among business users.
The AI code wars are heating up
OpenAI, Google, and Anthropic are leading the AI coding wars with tools like GitHub Copilot, Claude Code, and Codex, which can autocomplete code, build prototypes, and even replace human developers in some cases, potentially disrupting the software industry and changing the way we value software
Uncanny Valley: OpenAI and Musk Fight Again
The Uncanny Valley podcast discusses the ongoing feud between OpenAI and Elon Musk, as well as the Department of Justice's mishandling of voter data and the successful launch of the Artemis II mission. The podcast also touches on SpaceX's confidential IPO filing and the potential implications for the tech industry. |
|
Options & Tutorials
Building Hierarchical Agentic RAG Systems
This article introduces Protocol-H, a hierarchical agentic RAG system that bridges the modality gap between structured and unstructured data. It uses a supervisor-worker topology with autonomous error recovery, achieving 84.5% accuracy on the EntQA benchmark and reducing hallucination rates by 60%. The system is designed for enterprise-grade accuracy, safety, and auditability, making it suitable for production deployment.
Growing void between enterprise and frontier AI
The latest open weights AI models from Google, Microsoft, Alibaba, and Nvidia are becoming more capable and competitive, offering a cheaper alternative to frontier models for enterprises. These models can be run on relatively cheap hardware and don't require exposing sensitive customer data, making them an attractive option for companies.
Mozilla criticizes Microsoft's Copilot strategy
Mozilla's VP of global policy, Linda Griffin, criticized Microsoft's Copilot strategy, saying it prioritizes business over user choice. Microsoft had announced it would scale back some Copilot features in Windows after facing criticism for forcing AI on users without consent. Griffin argued that this move is too little, too late, and that Microsoft's tactics are degrading user choice and experiences on the web and in AI.
A Guide to Common AI Terms
This article provides a comprehensive guide to common AI terms, including AGI, AI agent, chain of thought, compute, deep learning, diffusion, distillation, fine-tuning, GAN, hallucination, inference, large language model, memory cache, neural network, RAMageddon, training, tokens, transfer learning, and weights.
Anthropic's Mythos Will Force a Cybersecurity Reckoning
Anthropic's new AI model, Mythos Preview, can discover vulnerabilities in software and develop exploits, sparking concerns about cybersecurity. Experts debate whether this is a true turning point or just AI hype, with some arguing it's a wake-up call for developers to prioritize security. |
|
Launches & Tools
Cloudflare Introduces EmDash: TypeScript CMS Positioned as WordPress Successor
Cloudflare introduces EmDash, a new open-source CMS built in TypeScript, powered by Astro 6.0, and designed as a spiritual successor to WordPress. EmDash aims to address security risks posed by WordPress plugins with secure sandboxes and explicit permissions, and includes AI-native features, developer tooling, and migration paths from WordPress.
Physical AI Device Shipments to Reach 145 Million Units by 2035
A study by Counterpoint Research predicts that physical AI device shipments will reach 145 million units by 2035, driven by growth in autonomous vehicles, robots, and drones. The report highlights the increasing demand for physical AI technology, with service robots and commercial drones expected to lead the growth. Humanoid robots are also expected to see significant growth, with shipments projected to exceed 100,000 units by 2028.
Google Chrome Rolls Out Protection Against Infostealers
Google Chrome has introduced a new feature called Device Bound Session Credentials (DBSC) to protect users from infostealers. DBSC associates authentication sessions with a specific device, making it difficult for attackers to steal session cookies. The feature is now available on Chrome 146 for Windows users and will be expanded to macOS in an upcoming release.
Linux 7.0 Release
Linux 7.0 has been released, featuring official Rust support, improved ARM and RISC-V support, and self-healing XFS. The release also includes code for ancient Alpha and SPARC CPUs. Linus Torvalds notes that AI tools are becoming increasingly useful in finding bugs, which may become the new normal for the kernel development process.
Netherlands First to Approve Tesla's Full Self-Driving
The Netherlands has become the first European country to approve Tesla's Full Self-Driving (Supervised) system. The country's regulatory authority, RDW, granted the approval after extensive testing, concluding that it is a positive contribution to road safety. However, the RDW noted that the driver remains responsible and must always remain in control. This approval marks Tesla's first regulatory green light for FSD use in Europe and could potentially lead to approval in other EU member states. |
|
Quick Links
Cryptographers place $5,000 bet on quantum computing impact
Two cryptographers, Filippo Valsorda and Matthew Green, have made a $5,000 bet on whether a quantum computer will be able to break current encryption methods. The bet is based on the ability of a quantum computer to recover a shared secret from a public key and ciphertext using either classical or quantum attacks. Valsorda believes that the transition to post-quantum cryptography needs to happen faster, while Green is skeptical about the immediate impact of quantum computing on cryptography.
New FCC rule will make Starlink satellite internet faster and cheaper
The FCC is voting on a new rule that will allow Starlink and other low-Earth orbit satellite systems to use more radio spectrum, resulting in faster internet speeds and potentially lower costs. The rule change will enable satellite operators to reach private agreements on how to share radio spectrum, allowing them to use more satellites in a given area or increase power levels per satellite.
TechCrunch Mobility: Who is poaching all the self-driving vehicle talent?
A talent war is brewing in the self-driving vehicle industry, with companies like Waymo, robotics, and defense tech firms competing for engineers with hybrid skills in classical robotics and AI. Base salaries are being pushed to $300,000-$500,000, and defense tech startups are leading the charge in compensation.
Slate Auto: Bezos-backed EV startup
Slate Auto is a Bezos-backed EV startup that aims to produce an ultra-cheap, customizable electric pickup truck. The company has been operating in secret for three years and has received funding from Jeff Bezos and other investors. Slate Auto's truck will have a starting price of around $25,000 and will be deeply customizable, with modular capabilities that allow owners to change the vehicle's configuration.
Onix: A Substack for AI Chatbots
Onix is a new company that offers a 'Substack for chatbots' where users can subscribe to AI versions of human experts in health and wellness. The platform uses 'Personal Intelligence' to protect user data and provides personalized advice, but raises concerns about the substitution of human interaction with AI models. |
Share NerdNewsShare your affiliate link to get commission!
Thanks for reading,
If you dont want to receive future editions of NerdNews, |