ZDNet's Steven Vaughan-Nichols shares his list of "what's new and improved" in the latest Linux 6.16 kernel. An anonymous reader shares an excerpt from the report: First, the Rust language is continuing to become more well-integrated into the kernel. At the top of my list is that the kernel now boasts Rust bindings for the driver core and PCI device subsystem. This approach will make it easier to add new Rust-based hardware drivers to Linux. Additionally, new Rust abstractions have been integrated into the Direct Rendering Manager (DRM), particularly for ioctl handling, file/GEM memory management, and driver/device infrastructure for major GPU vendors, such as AMD, Nvidia, and Intel. These changes should reduce vulnerabilities and optimize graphics performance. This will make gamers and AI/ML developers happier. Linux 6.16 also brings general improvements to Rust crate support. Crate is Rust's packaging format. This will make it easier to build, maintain, and integrate Rust kernel modules into the kernel. For those of you who still love C, don't worry. The vast majority of kernel code remains in C, and Rust is unlikely to replace C soon. In a decade, we may be telling another story. Beyond Rust, this latest release also comes with several major file system improvements. For starters, the XFS filesystem now supports large atomic writes. This capability means that large multi-block write operations are 'atomic,' meaning all blocks are updated or none. This enhances data integrity and prevents data write errors. This move is significant for companies that use XFS for databases and large-scale storage. Perhaps the most popular Linux file system, Ext4, is also getting many improvements. These boosts include faster commit paths, large folio support, and atomic multi-fsblock writes for bigalloc filesystems. What these improvements mean, if you're not a file-system nerd, is that we should see speedups of up to 37% for sequential I/O workloads. If your Linux laptop doubles as a music player, another nice new feature is that you can now stream your audio over USB even while the rest of your system is asleep. That capability's been available in Android for a while, but now it's part of mainline Linux. If security is a top priority for you, the 6.16 kernel now supports Intel Trusted Execution Technology (TXT) and Intel Trusted Domain Extensions (TDX). This addition, along with Linux's improved support for AMD Secure Encrypted Virtualization and Secure Memory Encryption (SEV-SNP), enables you to encrypt your software's memory in what's known as confidential computing. This feature improves cloud security by encrypting a user's virtual machine memory, meaning someone who cracks a cloud can't access your data. Linux 6.16 also delivers several chip-related upgrades. It introduces support for Intel's Advanced Performance Extensions (APX), doubling x86 general-purpose registers from 16 to 32 and boosting performance on next-gen CPUs like Lunar Lake and Granite Rapids Xeon. Additionally, the new CONFIG_X86_NATIVE_CPU option allows users to build processor-optimized kernels for greater efficiency. Support for Nvidia's AI-focused Blackwell GPUs has also been improved, and updates to TCP/IP with DMABUF help offload networking tasks to GPUs and accelerators. While these changes may go unnoticed by everyday users, high-performance systems will see gains and OpenVPN users may finally experience speeds that challenge WireGuard.

Read more of this story at Slashdot.

Jack Dorsey's new app Bitchat is now available on the iOS App Store. The decentralized, peer-to-peer messaging app uses Bluetooth mesh networks for encrypted, ephemeral chats without requiring accounts, servers, or internet access. Dorsey said he built it over a weekend and cautioned that it "has not received external security review and may contain vulnerabilities..." TechCrunch reports: The app's UX is very minimal. There is no log-in system, and you're immediately brought to an instant messaging box, where you can see what nearby users are saying (if anyone is actually around you and using the app) and set your display name, which can be changed at any time. [...] Dorsey has not directly addressed the fake Bitchat apps on the Google Play store, but he did repost another user's X post that said that Bitchat is not yet on Google Play, and to "beware of fakes."

Read more of this story at Slashdot.

Cisco has donated its AGNTCY initiative to the Linux Foundation, aiming to create an open-standard "Internet of Agents" to allow AI agents from different vendors to collaborate seamlessly. The project is backed by tech giants like Google Cloud, Dell, Oracle and Red Hat. "Without such an interoperable standard, companies have been rushing to build specialized AI agents," writes ZDNet's Steven Vaughan-Nichols. "These work in isolated silos that cannot work and play well with each other. This, in turn, makes them less useful for customers than they could be." From the report: AGNTCY was first open-sourced by Cisco in March 2025 and has since attracted support from over 75 companies. By moving it under the Linux Foundation's neutral governance, the hope is that everyone else will jump on the AGNTCY bandwagon, thus making it an industry-wide standard. The Linux Foundation has a long history of providing common ground for what otherwise might be contentious technology battles. The project provides a complete framework to solve the core challenges of multi-agent collaboration: - Agent Discovery: An Open Agent Schema Framework (OASF) acts like a "DNS for agents," allowing them to find and understand the capabilities of others. - Agent Identity: A system for cryptographically verifiable identities ensures agents can prove who they are and perform authorized actions securely across different vendors and organizations. - Agent Messaging: A protocol named Secure Low-latency Interactive Messaging (SLIM) is designed for the complex, multi-modal communication patterns of agents, with built-in support for human-in-the-loop interaction and quantum-safe security. - Agent Observability: A specialized monitoring framework provides visibility into complex, multi-agent workflows, which is crucial for debugging probabilistic AI systems. You may well ask, aren't there other emerging AI agency standards? You're right. There are. These include the Agent2Agent (A2A) protocol, which was also recently contributed to the Linux Foundation, and Anthropic's Model Context Protocol (MCP). AGNTCY will help agents using these protocols discover each other and communicate securely. In more detail, it looks like this: AGNTCY enables interoperability and collaboration in three primary ways: - Discovery: Agents using the A2A protocol and servers using MCP can be listed and found through AGNTCY's directories. This enables different agents to discover each other and understand their functions. - Messaging: A2A and MCP communications can be transported over SLIM, AGNTCY's messaging protocol designed for secure and efficient agent interaction. - Observability: The interactions between these different agents and protocols can be monitored using AGNTCY's observability software development kits (SDKs), which increase transparency and help with debugging complex workflows You can view AGNTCY's code and documentary on GitHub.

Read more of this story at Slashdot.

An anonymous reader quotes a report from Ars Technica: The rise of large language models like ChatGPT has led to widespread concern that "everyone is cheating their way through college," as a recent New York magazine article memorably put it. Now, OpenAI is rolling out a new "Study Mode" that it claims is less about providing answers or doing the work for students and more about helping them "build [a] deep understanding" of complex topics. Study Mode isn't a new ChatGPT model but a series of "custom system instructions" written for the LLM "in collaboration with teachers, scientists, and pedagogy experts to reflect a core set of behaviors that support deeper learning," OpenAI said. Instead of the usual summary of a subject that stock ChatGPT might give -- which one OpenAI employee likened to "a mini textbook chapter" -- Study Mode slowly rolls out new information in a "scaffolded" structure. The mode is designed to ask "guiding questions" in the Socratic style and to pause for periodic "knowledge checks" and personalized feedback to make sure the user understands before moving on. It's unknown how many students will use this guided learning tool instead of just asking ChatGPT to generate answers from the start. In an early hands-off demo attended by Ars Technica, Study Mode responded to a request to "teach me about game theory" by first asking about the user's overall familiarity with the subject and what they'll be using the information for. ChatGPT introduced a short overview of some core game theory concepts, then paused to ask a question before providing a relevant real-world example. In another example involving a classic "train traveling at speed" math problem, Study Mode resisted multiple simulated attempts by the frustrated "student" to simply ask for the answer and instead tried to gently redirect the conversation to how the available information could be used to generate that answer. An OpenAI representative told Ars that Study Mode will eventually provide direct solutions if asked repeatedly, but the default behavior is more tuned to a Socratic tutoring style. OpenAI said it drew inspiration for Study Mode from "power users" and collaborated with pedagogy experts and college students to help refine its responses. As for whether the mode can be trusted, OpenAI told Ars that "the risk of hallucination is lower with Study Mode because the model processes information in smaller chunks, calibrating along the way." The current Study Mode prompt does, however, result in some "inconsistent behavior and mistakes across conversations," the company warned.

Read more of this story at Slashdot.

skam240 writes: President Donald Trump's administration on Tuesday proposed revoking a scientific finding that has long been the central basis for U.S. action to regulate greenhouse gas emissions and fight climate change. The proposed Environmental Protection Agency rule would rescind a 2009 declaration that determined that carbon dioxide and other greenhouse gases endanger public health and welfare. The "endangerment finding" is the legal underpinning of a host of climate regulations under the Clean Air Act for motor vehicles, power plants and other pollution sources that are heating the planet.

Read more of this story at Slashdot.

Opera will file a complaint against Microsoft to Brazilian antitrust authority CADE on Tuesday, alleging the tech giant gives its Edge browser an unfair advantage over competitors. Opera claims Microsoft pre-installs Edge as the default browser across Windows devices and prevents rivals from competing on product merits. The company's general counsel Aaron McParlan said Microsoft locks browsers like Opera out of preinstallation opportunities and frustrates users' ability to download alternative browsers. Opera, which says it is Brazil's third-most popular PC browser, wants CADE to investigate Microsoft and demand concessions to ensure fair competition.

Read more of this story at Slashdot.

Google has admitted its earthquake early warning system failed to accurately alert people during Turkey's deadly quake of 2023. From a report: Ten million people within 98 miles of the epicentre could have been sent Google's highest level alert -- giving up to 35 seconds of warning to find safety. Instead, only 469 "Take Action" warnings were sent out for the first 7.8 magnitude quake. Google told the BBC half a million people were sent a lower level warning, which is designed for "light shaking", and does not alert users in the same prominent way. The tech giant previously told the BBC the system had "performed well" after an investigation in 2023. The alerts system is available in just under 100 countries -- and is described by Google as a "global safety net" often operating in countries with no other warning system. Google's system, named Android Earthquake Alerts (AEA), is run by the Silicon Valley firm - not individual countries.

Read more of this story at Slashdot.

Apple has lost its fourth AI researcher in a month to Meta [non-paywalled source], marking the latest setback to the iPhone maker's AI efforts. From a report: Bowen Zhang, a key multimodal AI researcher at Apple, left the company on Friday and is set to join Meta's recently formed superintelligence team, according to people familiar with the matter. Zhang was part of the Apple foundation models group, or AFM, which built the core technology behind the company's AI platform. Meta previously lured away the leader of the team, Ruoming Pang, with a compensation package valued at more than $200 million, Bloomberg News has reported. Two other researchers from that group -- Tom Gunter and Mark Lee -- also recently joined Meta. AFM is made up of several dozen engineers and researchers across Cupertino, California, and New York. In response to the job offers from Meta and others, Apple has been marginally increasing the pay of its AFM staffers, whether or not they've threatened to leave, said the people, who asked not to be identified because the moves are private. Still, the pay levels pale in comparison with those of rivals.

Read more of this story at Slashdot.