This week, we covered two unconfirmed Google search ranking updates, one mid-week and one over the weekend - some were pretty big. Google said Reddit gets no special preference in its search rankings or AI results...

Google Ads is rolling out a new customer acquisition option named "Report on new customers acquired." This lets you "Drive purchases from new and existing customers without changing bidding," according to Google.

Earlier this week, OpenAI sent an email to advertisers about new ChatGPT Ads Manager product updates. These updates include testing a new multi-product carousel format for product feed campaigns, plus a number of updates to the ad manager itself.

Google Ads will begin migrating or "upgrading" Automatically created assets and campaign-level broad match setting to AI Max starting on September 1, 2026. We knew this was coming and now the process is soon going to begin.

Gary Illyes from Google said that Google's crawlers do not only send HTTP requests such as GET and POST but also for HEAD, OPTIONS, PUT, PATCH, and DELETE. He said the latter make up for less than 1.5% of the total number of requests all Google's crawlers send. Google sends them for JavaScript, he said.

Google Ads added a "View call report" link in the Phone Call conversions leads report. This makes it easier to jump to the details of what makes up the metrics of that report and gives you access to that data in a new tab.

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.

Google Ads is now testing the "Strongest match" and "Strong match" labels it announced back in mid-June. This is the first time I have seen this label in the wild. I am actually a bit surprised Google went on with this test after the backlash...

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.