To serve the 397B-parameter Qwen 3.5 Mixture-of-Experts (MoE) model on Ironwood TPUs, engineers developed a modular JAX/Pallas optimization stack that achieved up to a 4.7x inference speedup for prefill-heavy workloads. The team bypassed severe hardware sharding constraints by deploying a hybrid Data Parallelism and Expert Parallelism (DP+EP) topology, paired with custom low-level communication fusions like a hierarchical reduce-scatter to optimize cross-device token routing. Finally, by executing hardware-aware custom kernels—such as Batched Ragged Page Attention and a fully-fused Gated DeltaNet (GDN) block—they successfully saturated HBM bandwidth and TensorCore MXUs to push system throughput near its theoretical roofline limits.

Conductor has evolved from a Gemini CLI extension into a portable plugin, bringing conversational Spec-Driven Development (SDD) to ecosystems like Antigravity CLI and Claude. Rather than relying on strict command sequences, developers can now chat naturally with their AI assistant while it dynamically manages persistent markdown artifacts (like spec.md and plan.md) in the background. This update eliminates workflow friction while ensuring your repository remains a version-controlled, single source of truth for your project's architecture and state across different AI tools.

Agent Platform's evaluation service is now generally available, providing developers with a unified engine to measure agent quality consistently across local development experiments and live production traffic. You can evaluate agents using over 20 pre-built metrics, DeepMind-backed adaptive rubrics, or custom code-based and LLM-as-a-judge metrics stored in a centralized, versioned registry. The service integrates directly into existing workflows via the Agent Platform SDK, agents-cli, and ADK, offering built-in user and environment simulators to automate complex multi-turn testing and streamline CI pipelines.

To prevent context window bloat and reduce token consumption, Genkit Go introduces Agent Skills based on a progressive disclosure architecture. Developers can package specialized instructions, scripts, and references into modular SKILL.md bundles where only the frontmatter metadata is initially exposed to the agent's system prompt. When a task matches the skill's description, Genkit's middleware dynamically loads the full instruction body and associated assets, ensuring the model accesses precise workflows exactly when needed.

Ray 2.55 introduces official, first-class support for Google Cloud TPUs, enabling developers to run distributed Python workloads on Google's accelerators using the familiar Ray task-and-actor APIs. To handle the strict networking requirement of keeping multi-host TPU "slices" together over their Inter-Chip Interconnect (ICI), the KubeRay Operator on GKE automatically provisions and labels the underlying hardware layout. Ray Core utilizes these labels via its slice_placement_group() primitive to atomically reserve complete slices, allowing developers to deploy jobs through KubeRay, Ray Train, or Ray Serve simply by declaring a hardware topology (like "4x4") without writing custom placement code.

To resolve the scaling bottlenecks and runtime errors caused by monolithic system prompts, engineering teams should treat prompts as build artifacts by modularizing instructions into reusable templates. By running these modular "skill files" through a transpiler, developers can enforce static validation, catch missing dependencies at build time, and integrate prompt generation directly into their CI/CD pipelines. This deterministic approach prevents code drift and ultimately establishes a safe framework where agents can propose updates to their own logic via standard pull requests.

Google Cloud has partnered with Parallel Web Systems to natively integrate Parallel's search infrastructure as a web grounding provider on the Gemini Enterprise Agent Platform. This integration enables developers to anchor their AI agents in verifiable, real-time web results, significantly improving factual accuracy for complex enterprise workflows. Additionally, the partnership offers expanded architectural flexibility, allowing users to programmatically extract, permanently cache, and process web data alongside other large language models.

We're excited to introduce LiteRT.js, the newest member of the LiteRT family! LiteRT.js is our powerful solution for running machine learning models directly in the browser, extending Google's cross-platform edge AI runtime to the web. Built for JavaScript developers, LiteRT.js delivers state-of-the-art ML model inference performance on WebGPU and upcoming WebNN, with a fallback to WebAssembly for CPU. This post provides a quick tour of LiteRT.js and gives web developers everything they need to get started.

Google Cloud API Gateway now offers a model routing feature in Public Preview, allowing developers to dynamically route traffic to models like Gemini, Claude, or OpenAI OSS-GPT without hardcoding endpoints or managing open-source proxies. Developers can easily configure these routing rules directly within their OpenAPI 3.x specifications by mapping virtual model names to specific backend targets on a shared host. Once deployed, the Gateway acts as a serverless ingress layer that accepts standard OpenAI-compatible requests, automatically transcodes the payload to the native schema of the target model, and routes the traffic on the fly.

Agent Plugins 1.0.0 is a new, vendor-neutral directory specification—backed by Google, Amazon, Microsoft, and others—for packaging Agent Skills and MCP servers into a single portable unit. By standardizing the manifest (plugin.json) and utilizing a fixed directory layout, it eliminates the need for developers to maintain separate wrappers or configurations to support different AI coding agents and IDEs. Google has officially joined as a Core Maintainer and already rolled out support in the Agents CLI and Data Agent Kit, allowing developers to start building and distributing interoperable plugins today.

WordPress 7.0.3 is now available

WordPress 7.0.3 is now available which features several security fixes. Because this is a security release, it is recommended that you update your sites immediately.

You can update to WordPress 7.0.3 by downloading it from WordPress.org, or visiting your site’s Dashboard → Updates and clicking Update Now. Sites that support automatic background updates will begin updating shortly.

For more information, please visit the WordPress 7.0.3 HelpHub site.

Security updates included in this release

The security team would like to thank the following people for responsibly reporting vulnerabilities and allowing them to be fixed in this release:

• Pre-auth reflected cross-site scripting (XSS) on the login screen with potential to lead to PHP code execution reported by the team at pwn.ai.
• Contributor+ stored cross-site scripting (XSS) in posts via the emoji settings element reported by Asaf Mozes (amosec)
• Contributor+ stored cross-site scripting (XSS) in the Post Content block reported by n05ec
• Contributor+ stored cross-site scripting (XSS) in Quick Edit on sites with a large number of users reported by Naveen S and Ajmal Moochingal
• Contributor+ stored cross-site scripting (XSS) in the Post Date block reported by Alex Concha of the WordPress Security Team
• A privilege escalation issue on multisite networks with user registration enabled, allowing a user to create a new site reported by Aikido Security
• An information disclosure issue in the Latest Comments block exposing comments on password-protected posts reported by Ehtisham Siddiqui of the WordPress Security Team
• Enumeration of post slugs reported by HDWSec
• Disclosure of notes in comment feeds reported by Elio Gubser
• Author+ CSS injection via a bypass of the safe CSS attribute filter reported by Anthropic
• Bypass of the email address confirmation flow reported by 0ways
• A server-side request forgery (SSRF) issue in URL validation allowing requests to link-local ranges reported by Andrew Mohawk and multiple independent reporters

Backports

As a courtesy, these fixes are being backported, where necessary, to all branches eligible to receive security fixes (currently through 4.7). As a reminder, only the most recent version of WordPress is actively supported. The backports are in progress and will ship as they become ready.

WordPress 7.1 RC2 has also been released, containing all applicable fixes.

CVE and GHSA references

Details of the login screen XSS vulnerability can be found in the advisory: CVE-2026-64638 / GHSA-52p2-r8wf-jcrf.

Thank you to these WordPress contributors

This release was led by John Blackbourn. In addition to the security researchers mentioned above, WordPress 7.0.3 and its backports would not have been possible without the significant contributions of the following people:
Aaron D. Campbell, Aaron Jorbin, Adam Silverstein, adrianmoldovanwp, Aki Hamano, Alex Concha, Andrew Duthie, Andrew Serong, annezazu, Barry, Bernie Reiter, Daniel, Daniel Richards, David Biňovec, Dennis Snell, Ehtisham Siddiqui, Erwan Le Rousseau, Fabian Kaegy, fiocavallari, George Mamadashvili, gubser, Isabel Brison, Jarda Snajdr, Jb Audras, Jeremy Felt, Joe Dolson, Joe Hoyle, John Blackbourn, Jon Surrell, Jonathan Desrosiers, Khokan Sardar, Lance Willett, lucasbustamante, lucatume, Marco Ciampini, Marin Atanasov, Mohammad Jangda, Mukesh Panchal, Paul Kevan, Peter Wilson, ramonopoly, SergeyBiryukov, vortfu, Weston Ruter

Join us for the launch of WordPress 7.1 at WordCamp US 2026, August 16–19.

Vietnam’s FPT says it has joined the OpenAI Partner Network as a Select Partner, with plans to expand enterprise AI and cybersecurity services across Asia-Pacific. The announcement gives APAC organizations another potential implementation option but leaves questions about customer deployments, performance, and data governance.

The post Vietnam’s FPT Says It Joined OpenAI Partner Network to Expand Enterprise AI in APAC appeared first on TechRepublic.

Developer and creator Tom Shaw sits down with Meta’s VP of Data Centers Rachel Peterson to discuss how Meta builds the infrastructure that powers Instagram, Facebook, WhatsApp, Threads, Meta AI, and more.

Listen as they discuss why Meta designs and operates its own custom data centers instead of relying on third-party infrastructure, how Meta is preparing for the future of AI, and the engineering challenges involved in building some of the world’s largest computing facilities.

They’ll answer questions such as:

• Why does Meta build its own AI data centers?
• What does the infrastructure behind Meta AI actually look like?
• How are AI data centers cooled efficiently?
• How much water do Meta’s data centers really use?
• How does Meta choose where to build new data centers?
• How does Meta’s infrastructure support personal superintelligence?

The post Why Meta Builds Its Own AI Data Centers appeared first on Meta Newsroom.

ChatGPT Atlas is scheduled to stop working Aug. 9, 2026. Users should export browser data, verify account access, and test replacement workflows before uninstalling it.

The post ChatGPT Atlas Shuts Down Aug. 9: What Users Must Save Before Migrating appeared first on TechRepublic.

The Django Software Foundation is announcing a call for an Executive Director. The Executive Director is the operational leader of the Django Software Foundation, a paid position reporting to the Board of Directors, responsible for setting the Foundation's strategic direction and turning it into action, while managing day-to-day operations. They serve as the main connector between the Board, staff, community, and sponsors.

The Django Software Foundation (DSF) is a 501(c)(3) nonprofit that develops and maintains Django, a free and open-source web application framework. The Foundation exists to support the development of Django by sponsoring sprints, meetups, gatherings and community events; to promote the use of Django among the web development community; to protect the framework's intellectual property and long-term viability; and to advance the state of the art in web development.

This is a new role for the Foundation. Django itself has been around since 2005, but the DSF wasn't founded until 2008, and the person who takes on this role will play a key part in maturing the Foundation's internal structure, helping ensure the DSF can properly support and sustain this important ecosystem going forward. The position is initially for a period of one year, renewable subject to an annual performance evaluation. Depending on the candidate, the role may be part-time or full-time.

Beyond running the Foundation, the Executive Director is a representative of the DSF itself. They embody Django's welcoming culture and help the community sustain the framework's home. The Executive Director is often called upon to represent the Foundation publicly, including at Django conferences and events, and to grow awareness of the DSF as an organization, distinct from the framework it supports.

Responsibilities

Executive Director duties include (but are not limited to):

• Fundraising: leading sponsorship development, corporate and individual membership growth, and partner relationships, including support for the jump from our current 300K USD annual fundraising goal to 500K USD. At the current funding level (around 300K per year), a full-time Executive Director isn't yet sustainable. We'd like to fix that, and we want you to lead that change.
• Admin and operations management: day-to-day operations and administration of the DSF, financial reporting, grant management, and the general running of the organization. Over time, helping grow the DSF into a more mature organization by establishing the operational foundations that support the nonprofit's growth.
• Managing the DSF Assistant and Fellows: overseeing the DSF Assistant and the Django Fellows program, the paid maintainers funded by the DSF.
• Marketing and outreach: community outreach and communications, representing the DSF publicly (for example, conference representation), and growing awareness of the Foundation as distinct from the framework.
• Legal, trademark, and follow-ups: overseeing international trademark policy enforcement, creating, signing, and renewing contracts, handling legal correspondence, and the unglamorous administrative follow-through that keeps a 501(c)(3) compliant. First-hand legal knowledge isn't required here; you'll work with counsel.
• Working group check-ins: regular coordination with the DSF working groups, the volunteer committees handling events, AI, accessibility, fundraising, and more.
• Working with our Django events and conferences like our DjangoCons.

Requirements

An Executive Director is responsible for fundraising, operations, communications, and community coordination. This is a broad remit, and it isn't our expectation that you come into the job an expert in every part of it. We hope you'll have solid experience in a few of these areas, particularly the ones most central to the role (fundraising and partnership development, nonprofit operations, and stakeholder communication). A willingness to learn and a demonstrated history of doing so are more important than comprehensive knowledge.

The areas you can expect to work across include (and are not limited to):

• Fundraising, sponsorship, and partnership development
• Nonprofit operations, financial reporting, and grant management
• Contracts, trademark, and 501(c)(3) compliance (in coordination with counsel)
• Public representation, marketing, and communications
• Coordinating staff, volunteers, and working groups
• Technical knowledge is not required, but is a nice-to-have:
• Knowledge of, or familiarity with, the Django and Python community
• Familiarity with open source licenses and communities

And required professional skills such as:

• Conflict resolution
• Time management and prioritization expertise
• Ability to focus in short periods of time and do substantial context switches
• Self-awareness to recognize their own limits and reach out for help
• Relationship-building and coordination with the Board, staff, working groups, sponsors, and external parties
• Tenacity, patience, compassion and empathy are essential

Therefore, a Django Executive Director requires the skills and judgment of an experienced nonprofit leader who is comfortable with fundraising, operations, and coordination with community stakeholders. Open-source experience and familiarity with the Django or Python community in particular are a big plus.

Being part of the Django community isn't a prerequisite for this position. We'll consider applications from anyone with a proven history of nonprofit leadership or comparable experience in an open-source or mission-driven community, but this is a remote position based in the United States, and unfortunately we are not able to offer visa sponsorship for this role.

The DSF is an equal opportunity employer. We welcome applicants of every background and don't discriminate on the basis of race, color, religion, gender, gender identity or expression, sexual orientation, national origin, disability, age, or veteran status.

How to apply

If you're interested in applying for the position, please submit your application via hiring@djangoproject.com. Your application should include:

• A cover letter (optional)
• A resume or CV
• A brief vision statement (500 to 1000 words) addressing your understanding of the Foundation's current position, the key opportunities and challenges you see for the Foundation, and your approach to the role

References may be requested during the interview process.

The compensation for this role is a base salary of $90,000 to $120,000, plus a bonus of up to $60,000 tied to our progress toward the $500,000 fundraising goal, which we'll tier as we work toward it. Depending on the candidate, the DSF will consider a part-time position and adjust the salary accordingly.

Applicants will be evaluated based on the following criteria:

• Relevant nonprofit leadership and operational experience
• Track record in fundraising and partnership development
• Understanding of the position and of the DSF's current stage
• Clarity, formality, and precision of communications
• Familiarity with open source and/or the Django and Python community
• Strength of reference(s)

Applications will be open until midnight Central Time, September 14, 2026, with the expectation that the successful candidate will start around November 1, 2026 (to be confirmed).

Reference: Announcing the Search for a DSF Executive Director (Django Project blog, June 17, 2026).

Here is a recap of what happened in the search forums today, through the eyes of the Search Engine Roundtable and other search forums on the web. We are seeing big changes and swings with the Google Search results...

Google's August Android 17 update fixes Pixel 10 gaming, GPU, and touchscreen issues while bringing the latest security patch to supported Pixel devices.

The post Google Rolls Out August Android 17 Update With Pixel 10 Stability Fixes appeared first on TechRepublic.

This second installment explores how Ray’s higher-level libraries—Serve, Data, and Train—abstract the complexities of running AI workloads on Google's TPU slices. Ray Serve uses a simple topology configuration to correctly gang-schedule large multi-host models, while Ray Data eliminates data-loading bottlenecks by feeding accelerators directly with native JAX batches. Finally, JaxTrainer streamlines distributed training across TPUs by automatically handling cross-slice coordination, checkpointing, and fault tolerance.

Building AI agents often leaves developers uncertain if prompt tweaks to fix single errors will accidentally cause widespread regressions in production. To bridge this gap, Google has introduced a new developer skill for coding agents that automates a five-stage evaluation flywheel: preparing data, running inference, grading with adaptive AutoRaters, analyzing failure clusters, and executing targeted optimizations. Running continuously against production traffic or on-demand via synthetic scenarios, this tool allows developers to describe testing goals in plain language while an independent evaluation service safely validates and counts actual performance improvements.

Samsung’s Galaxy Z Fold8 leads record foldable preorders in the US and South Korea as more buyers move from Flip models to the larger Fold lineup before launch.

The post Samsung: New Galaxy Foldables Beat Previous Preorder Record by 30% appeared first on TechRepublic.