AI Tech News Past 24 Hours

Security researchers have uncovered a new zero-click exploit chain for the Google Pixel 10 that grants attackers full root access. Building on similar findings from the Pixel 9, this demonstrates that modern Android devices can still be completely compromised without any user interaction.

Codex is now available in the ChatGPT mobile app, enabling users to manage coding tasks on the go. This update allows developers to monitor, steer, and approve these processes in real time across various devices and remote environments.

A recent tech commentary warns that relying on traditional machine learning techniques like the classic sigmoid activation function is no longer sufficient for solving complex modern artificial intelligence challenges. The piece argues that developers and researchers must move beyond these outdated foundational models to embrace more advanced neural network architectures. Ultimately, it serves as a reality check for the industry to stop overestimating the capabilities of legacy algorithms in the current AI boom.

The U.S. Department of Justice is demanding that Apple and Google reveal the personal data of over 100,000 users who downloaded the EZ Lynk Auto Agent car-tinkering app. Federal authorities are seeking to unmask these drivers in order to investigate the use of the vehicle modification software.

A new interactive project allows users to watch a neural network train itself to play the classic game Snake. While the AI can achieve near-perfect scores on a small grid, observers note that the model's performance eventually collapses. Users speculate this degradation is caused by numerical instability, flawed reward structures, or the snake getting trapped in an endless loop when it runs out of space.

The preprint server arXiv has implemented a new policy that issues a one-year ban for authors who submit papers containing hallucinated references generated by AI. The platform emphasized that under its Code of Conduct, authors remain fully responsible for all content within their submissions regardless of how the material was generated.

Digital archivist and tech historian Jason Scott maintains the weblog ASCII to document computing history, internet culture, and digital preservation efforts. The site serves as a personal platform for his extensive research and ongoing archival projects. Readers can expect insightful commentary on the evolving technological landscape and the importance of preserving early digital media.

A new open-source GitHub project called Image-blaster introduces an image-to-world skillset for the Claude AI model. The tool enables users to instantly generate detailed 3D environments, meshes, and sound effects from just a single input image. Developers can access and contribute to the ongoing development of the project directly on GitHub.

DPhi Space's ClusterGate-2, a pure-OCaml CCSDS protocol stack featuring end-to-end-encrypted command and control and post-quantum key rotation, successfully booted in orbit aboard the Borealis satellite. This milestone establishes a highly secure foundation for space communications, with the next iteration of the technology already planned as OxCaml.

Start9 is expanding into hardware with a new RISC-V router powered by StartWRT, a customized fork of OpenWrt featuring a modern, simplified interface designed for non-technical users. Community members express mixed reactions, praising the accessible design but raising concerns about the startup's hardware and software risks, high pricing, and the need for real-world performance benchmarks. Furthermore, power users debate the merits of a full software fork over upstream contributions and worry that the simplified GUI might restrict access to advanced networking configurations.

The UK government has released new details and photographs of an airdrop mission to the remote island of Tristan da Cunha. Executed on 9 May 2026, the intrepid operation successfully delivered urgent medical support to the isolated territory. This logistical feat highlights the complex planning required to provide emergency aid to one of the world's most inaccessible locations.

Ontario auditors discovered that AI medical note-taking systems used by doctors frequently generate inaccurate patient records. Evaluations revealed that 60% of the tested AI scribe tools incorrectly mixed up prescribed medications in their notes. These errors raise significant concerns about the reliability and safety of deploying artificial intelligence in healthcare settings.

Waymo has initiated a voluntary recall of approximately 3,800 robotaxis to address a software flaw that could cause the vehicles to drive into flooded roadways. The company is deploying an over-the-air update to ensure the autonomous cars safely avoid navigating through deep standing water.

Amazon and other big tech companies are pressuring employees to increase their AI tool usage as a performance metric, prompting workers to fabricate tasks just to burn through AI tokens. Critics argue this creates a perverse incentive that inflates token consumption without actually improving real productivity or development velocity.

The Open Vehicle Monitoring System project is gaining traction as electric vehicle owners seek to bypass restrictive OEM CAN gateway modules that limit standard OBD-II ports to read-only access. While manufacturers claim these lockouts improve security, critics argue they are primarily designed to monetize telemetry data and force users into proprietary software subscriptions. Consequently, drivers are increasingly turning to open-source tools and hardware workarounds to regain full diagnostic control and privacy for their vehicles.

Access to advanced frontier AI is expected to become increasingly scarce and selective in the near future due to mounting economic and security constraints. As development costs soar and safety concerns grow, only a limited group of well-resourced organizations will be able to leverage these powerful technologies.

A vehicle owner physically removed the modem and GPS from their 2024 RAV4 Hybrid to prevent the car from transmitting continuous personal telemetry. This DIY privacy modification highlights growing consumer pushback against modern vehicles functioning as rolling data trackers.

A recent hardware experiment successfully connected an Nvidia RTX 5090 desktop GPU to an M4 MacBook Air using an external enclosure. The extreme setup tests whether Apple's ultraportable laptop can handle high-end gaming when paired with top-tier desktop graphics hardware.

Despite Apple spending five years fortifying its systems against memory corruption, security researchers have developed the first public macOS kernel exploit for the new M5 chip. Working alongside Mythos Preview, an engineering team successfully built a functional exploit in just five days.

A newly discovered exploit for the Nginx web server is now available on GitHub, specifically targeting the CVE-2026-42945 vulnerability. Developed by DepthFirstDisclosures, the exploit tool named Nginx-Rift is currently open for community code review and contributions.

The newly launched Claude for Small Business package integrates the AI assistant directly into the everyday tools used by smaller enterprises. This offering includes a suite of connectors and ready-to-run workflows designed to streamline daily operations.

The UK government has replaced its use of Palantir software with a newly developed, internally built IT system for managing refugee data. Officials stated that the in-house solution offers greater flexibility while maintaining high security standards.

The York University Computer Museum (YUCoM) serves as a historical collection and research center dedicated to the history of computing in Canada. By focusing on the Canadian computer hobby movement, the museum preserves the nation's early technological milestones. It remains a vital resource for understanding the evolution and heritage of domestic computing.

This piece breaks down the GGUF file format to explain the essential metadata and extra components needed to run a language model independently of its raw weights. Additionally, it highlights the current limitations of the format by identifying what crucial information is still missing for seamless deployment.

A new open-source project introduces a specialized skill for AI coding assistants like Claude Code and OpenAI Codex to foster deliberate practice. The tool, named learning-opportunities, helps developers actively refine their programming abilities while working alongside AI. By prompting users to engage critically with the code, it prevents over-reliance on automated generation and promotes continuous learning.

An author reflects on the overwhelming temptation to use artificial intelligence for everyday writing tasks such as drafting articles, documents, and code. They warn that this growing dependency on automated tools is actively diminishing their personal writing skills and cognitive sharpness.

Anthropic has released an open-source GitHub repository called Claude for Legal, providing a suite of plugins tailored specifically to streamline legal workflows. Developers are encouraged to contribute to the project's ongoing development to help integrate the company's AI capabilities into specialized legal processes.

A new open-source project named Nibble allows developers to generate LLVM IR without using malloc or relying on external dependencies. The lightweight tool is available on GitHub and is designed to be entirely self-contained. This makes it a highly practical solution for low-level code generation in constrained environments.

Driven by severe drought in the Plains, the USDA predicts the US wheat harvest will drop to its lowest level since 1972 for the 2026/27 season. In contrast, the agency forecasts record soybean production gains as farmers navigate ongoing challenges from trade tensions and rising input costs.

Meta is testing a new Threads feature that allows users to tag a Meta AI account to ask questions and receive answers. However, users do not have the option to block the AI account if they prefer not to interact with it.