AI platforms

Google has filed an appeal against European Union regulatory requests that would require it to open up Android to rival AI assistants and share search data with competitors. The case adds to Google's long-running battles with EU antitrust authorities, and is being closely watched as regulators across Europe push to curb the power of large technology platforms in emerging AI markets.

Nvidia has unveiled a new security platform designed to prevent autonomous AI agents from acting outside their intended instructions. The company says the tooling aims to monitor and control agent behaviour as businesses deploy AI systems that operate independently. Coverage from wire and business outlets highlights growing industry concern about AI safety and oversight as agentic AI adoption accelerates.

Chinese AI startup Moonshot is reportedly seeking access to additional Nvidia Blackwell chips, according to a report circulating on market discussion platforms. The claim comes just a week after a Trump administration official flagged an alleged breach of US export controls on advanced chips. The report highlights ongoing tensions over US restrictions on shipping cutting-edge semiconductors to Chinese firms, a flashpoint in the tech rivalry between Washington and Beijing.

A new site called Agent.reviews has launched, positioning itself as a platform where AI agents both read and write reviews of software tools. The project was shared with the Hacker News community, drawing moderate engagement. It taps into growing interest in agentic AI workflows, where automated agents research and evaluate developer tools on behalf of users.

A new open-source project called OpenChart is being presented as a free alternative to TradingView, a popular charting platform for traders, with the added feature of a built-in AI agent. The project is hosted on GitHub, where developers can inspect and contribute to the code. Interest is centred on whether an open-source tool can match established trading chart software while offering AI assistance.

A writer has marked eleven instalments of an ongoing essay series on dependence in modern software, arguing that whoever controls a layer you don't own ultimately controls you. The series has examined layers that are effectively rented rather than owned: distribution, the model, identity, access, the harness, the meter and the runtime. The thread ties together concerns about platform dependence, shifting pricing and vendors changing terms under builders who rely on them.

A TypeScript project called context-mode offers context window optimization for AI coding agents. It claims to sandbox tool output with a 98% reduction, persist session memory between tasks, and enforce routing across 17 supported platforms via MCP and hooks. Developers interested in making coding agents cheaper and more reliable are taking note.

Nvidia has introduced an Open Agent Safety Platform, a reference framework for continuous in-silicon monitoring of AI agents. Announced on the company's developer blog, the platform is aimed at helping developers keep autonomous agents safe and observable directly at the hardware level. Early reaction among developers is focused on what the reference design means for deploying agents in production.

Figma has restricted access to its Model Context Protocol (MCP) server to a whitelist of approved clients, and Pi is not among them. The move means developers relying on Pi to connect with Figma through MCP can no longer do so, prompting discussion about platform openness and control over AI tooling.

A GitHub repository called andrej-karpathy-skills is drawing attention. It packages a single CLAUDE.md configuration file intended to improve Claude Code's behaviour, built from observations Andrej Karpathy has shared about common pitfalls of LLM-assisted coding. The project has reached the platform's trending list globally, with early engagement from developers interested in tuning AI coding assistants.

Joe Tsai, chairman of Alibaba, has urged Europe to embrace open source technology as its only viable route to achieving technological independence. His remarks come amid ongoing debate in Europe about reliance on American and Chinese tech platforms and the region's push to build its own digital capabilities. Tsai's comments add to the wider discussion about how Europe can compete in AI and software without depending on foreign proprietary providers.

A tool called Hermes Agent is being presented as a way for people who use several AI assistants to manage their workflows from one place. The pitch is that users juggling multiple AI services no longer need to switch between them, instead coordinating tasks through a single agent. Discussion is focused on whether this kind of consolidation can genuinely streamline work across different AI platforms.

Orkut Buyukkokten, creator of the once-hugely popular social network Orkut, is working to relaunch the platform. The new version is being positioned against algorithmic feeds and AI-generated content, with the founder arguing for a return to more human, authentic online socialising. Orkut was shut down by Google in 2014 but retains nostalgic appeal, particularly in Brazil and India, and the revival has drawn interest from former users and tech commentators.

Russian tech outlets report that 'vibe coders' — programmers using AI-assisted, prompt-driven development — have recreated a version of Minecraft that runs inside Twitter, with hundreds of thousands of users reportedly trying it out. The project is drawing attention for showing how far AI coding tools and browser-based games can be pushed inside social platforms, and reaction online has focused on the novelty of playing a full game from a social feed.

Temporal has published a piece on running the Pi coding agent on its workflow orchestration platform, framing it as 'the immortal life of Pi'. The post explains how durable execution keeps long-running coding agents alive across failures and restarts, a growing concern as AI agents take on extended, multi-step development tasks.

Upscale AI, a startup backed by Nvidia, has launched a platform designed to connect chips from competing suppliers, according to Reuters. The move signals an effort to make different chip vendors interoperable in AI data centres, potentially reducing reliance on any single supplier. It is attracting attention as a notable challenge to the dominance of integrated chip ecosystems in the fast-growing AI hardware market.

Gigabyte's W775-V10-L01 is a deskside AI workstation built around Nvidia's GB300 platform, giving developers and researchers supercomputer-class AI hardware outside the data center. A hands-on review has drawn attention to its size, power requirements and price, with readers debating who would run such a machine on or under a desk rather than renting cloud capacity.

SK hynix and Samsung are moving ahead with 16-layer HBM4 high-bandwidth memory technology, while Nvidia's upcoming Vera Rubin AI platform is expected to rely largely on 12-layer HBM4 chips, according to The Korea Herald. The development underscores fierce competition among Korean memory makers to supply advanced memory for Nvidia's next generation of artificial intelligence accelerators.

An Ethereum researcher has urged blockchain developers to adopt what they call 'bunker mode' — a hardened security posture designed to protect networks and crypto users from risks posed by increasingly capable AI systems. The warning centers on the potential for AI to automate scams, exploit vulnerabilities and manipulate markets at unprecedented speed. Crypto community reactions are mixed, with some backing stronger defensive coordination and others questioning how practical such measures would be across decentralized networks.

A writer has expanded a short social media comment into a full article on Substack about a child safety law that, in his view, treats certain groups unfairly. The piece touches on gaming, retrocomputing and the EU's approach to online safety and AI-related regulation. Reaction so far is small but the topic sits at the centre of ongoing debate in Europe about protecting children online without restricting access to games and computing platforms.

A new podcast conversation between journalist Alex Heath and sources close to the AI industry takes up one of tech's biggest open questions: who will control and profit from artificial intelligence in the years ahead. The discussion weighs the power of major labs, big tech platforms and open-source efforts, and the conversation has drawn tens of thousands of engagements from viewers following the industry's ownership and governance battles.

J.P. Morgan examines how embedded finance, the integration of financial services into non-financial platforms, is adapting to counter fraud increasingly powered by artificial intelligence. As AI tools make scams faster and more convincing, firms are embedding stronger verification, monitoring and security measures directly into financial products to protect transactions across digital ecosystems.