Tech News: Latest Updates 2026-03-25

This report is a summary of technology news from 2026-03-25 and has been prepared solely based on the original source texts.

This report was automatically generated by AnishoAI. For automating your website’s content production, visit anishoai.com for more information and to view plans.

Copilot: A Tool for Higher-Value Work at KPMG Canada

According to a report by news.microsoft.com, Christine Andrew, Managing Director for AI Enablement at KPMG in Canada, believes that the primary value of Microsoft 365 Copilot is not merely saving time, but enabling a focus on more strategic and higher-quality activities. According to her, before using this tool, a large portion of management time was spent on planning, execution, and reporting, but now with the reduction of this operational burden, she has more time for deep thinking about market trends, risks, and client needs.

Andrew heavily relies on Copilot in Teams to manage meetings that are often scheduled two or three at a time. She also uses Copilot Chat as a “thought challenger” to review her draft strategies and see how a CEO with a different perspective might react. In preparing a recent keynote, she also used Copilot’s research capabilities to analyze the audience, which resulted in a re-invitation from the conference organizers.

According to this report, Christine Andrew launched the early-access Copilot program at KPMG Canada about three years ago with just 40 licenses. This number later grew to one thousand users and was ultimately deployed to all 12,000 employees in 2025. According to her, Copilot’s deep integration with the Microsoft ecosystem has made it easier and safer to use compared to other AI tools. Today, usage is tracked weekly, employees are required to set AI-related goals, and numerous training programs have been conducted.

According to Andrew, the goal of expanding Copilot was not merely to follow new technology, but to free people from low-value tasks so they would have more time for thinking, advising clients, and solving more complex problems. She emphasizes that Copilot does not shorten the workday, but rather changes the type of activities performed during that time.

Amazon Timestream for InfluxDB Now Available in Three New AWS Regions

According to an announcement by aws.amazon.com, the Amazon Timestream for InfluxDB service is now available in the Mexico (Central), Japan (Osaka), and Brazil (Sao Paulo) AWS Regions. This service allows developers and DevOps teams to run fully managed InfluxDB databases on AWS for real-time and time-series applications using open-source APIs.

Timestream for InfluxDB offers capabilities including Multi-AZ high availability, read replicas, improved durability, and multi-node scalability. According to the source announcement, users can start with a single-node configuration and scale up to a 15-node Enterprise cluster without needing to re-architect their infrastructure.

Users can create their InfluxDB databases through the Amazon Timestream for InfluxDB console, AWS CLI, or AWS SDKs. For more information, documentation and the pricing page for this service are available.

Cloudflare Introduces New Dynamic Workers Sandbox

According to an announcement by blog.cloudflare.com, Cloudflare has released a new capability called Dynamic Worker Loader in public beta for all paid Cloudflare Workers users. This capability allows a Cloudflare Worker to launch a new Worker at runtime in its own dedicated sandbox with code generated on-the-fly. Previously in September, the company had introduced the concept of Code Mode and demonstrated that converting an MCP server to a TypeScript API could reduce token consumption by up to 81%.

According to this report, Dynamic Workers use a sandboxing mechanism called isolate — the same JavaScript V8 execution engine used in Google Chrome. The company states that this technology starts up approximately 100x faster than containers and is 10x-100x more memory efficient.

Among other advantages of Dynamic Workers are the absence of limits on the number of concurrent sandboxes and their creation rate. The service is supported across all hundreds of Cloudflare locations worldwide, and one-off Workers typically run on the same machine and even the same thread that created them. In terms of programming language, Dynamic Workers support JavaScript, and according to the documentation, Python and WebAssembly are also possible, but for short AI-generated code, JavaScript loads and executes faster.

Auto-Dismiss Policies in GitLab for Vulnerability Management

According to an announcement by about.gitlab.com, the auto-dismiss vulnerability policies capability in GitLab enables automatic management of unnecessary security findings. Security scanners often produce findings from test code, vendored dependencies, auto-generated files, and known false positives that bury real vulnerabilities beneath them. Security teams spend hours manually dismissing repetitive and irrelevant findings, leading to alert fatigue and reduced adoption of security scanning.

With this capability, criteria can be defined based on file path, directory, or vulnerability identifier (CVE or CWE) and recorded in a YAML file. Policies are automatically applied on every default-branch pipeline. According to the documentation, up to 1,000 vulnerabilities are processed per run. Auto-dismissed findings include a documented reason and a link to the relevant policy, and unlike scanner exclusions, they remain in the report so they can be reviewed if conditions change.

As an example, a policy can be defined to automatically dismiss vulnerability classes such as XSS with identifier CWE-79 or SQL injection with identifier CWE-89 that are covered by a WAF. Additionally, a sample is provided for accepting risk for the Log4j CVE family with the pattern CVE-2021-44*, titled “Accept risk for log4j CVE family” with the description “Log4j CVEs mitigated by version pinning and WAF.” This capability is accessible from Secure > Policies > New policy > Vulnerability management policy and is activated by merging the MR.

Power-Flexible AI Factories for Stabilizing the Power Grid

According to a report by NVIDIA, Emerald AI, in collaboration with NVIDIA, EPRI, National Grid, and Nebius, has demonstrated that “power-flexible” AI factories can automatically reduce their power consumption during peak demand. This technology could enable faster connection of these facilities to the power grid without the need for long-term infrastructure upgrades and prevent electricity rate increases for regular consumers.

Following initial tests in Arizona, Virginia, and Illinois, the Emerald AI team in December deployed the Emerald AI Conductor Platform to Nebius AI’s new AI factory in London. At this facility, real workloads ran on a cluster of 96 NVIDIA Blackwell Ultra GPUs connected via the NVIDIA Quantum-X800 InfiniBand platform. EPRI and National Grid simulated various stress scenarios, including recreating the energy consumption surge from the UEFA EURO 2020 football match between England and Germany. In that actual event, National Grid, which supplies electricity to England and Wales, experienced a demand surge of approximately 1 gigawatt within minutes.

According to this report, the AI cluster successfully reduced its power consumption and acted as a buffer for the sudden energy surge without disrupting high-priority tasks. Emerald AI, according to its announcement, recorded 100% compliance with more than 200 energy consumption targets set by EPRI and National Grid. Varun Sivaram, founder and CEO of Emerald AI, said: “With this technology, AI factories become useful assets for the power grid.” Steve Smith, Chief Strategy Officer at National Grid, also stated that the tests went beyond those conducted in the United States, and in addition to GPUs, CPUs and the total power consumption of IT equipment were also tested. After four successful tests, Emerald AI and NVIDIA are ready for real-world deployment of this technology at the Aurora AI Factory in Virginia.

Summary

Today’s main trends include the expansion of AI tools in the workplace for focusing on more strategic activities, the development of cloud and security infrastructure with automation capabilities, and innovation in energy consumption management of AI data centers. Companies such as Microsoft, Cloudflare, GitLab, and NVIDIA are simultaneously pursuing organizational productivity and infrastructure sustainability by offering new solutions.

Frequently Asked Questions

What impact has Microsoft 365 Copilot had on how managers work at KPMG Canada?

According to Christine Andrew, Copilot has reduced the operational burden of planning, execution, and reporting, providing more time for deep thinking about market trends, risks, and client needs. Additionally, Copilot in Teams is used for managing concurrent meetings, and Copilot Chat is used as a thought challenger.

What is the Dynamic Worker Loader capability in Cloudflare and how does it work?

This capability, released in public beta for paid Cloudflare Workers users, allows a Worker to launch a new Worker at runtime in a dedicated sandbox with code generated on-the-fly. This sandbox uses the isolate mechanism based on Google Chrome’s JavaScript V8 engine.

What problem do auto-dismiss policies in GitLab solve?

This capability solves the alert fatigue problem for security teams. By defining criteria based on file path, directory, or vulnerability identifier (CVE or CWE) in a YAML file, unnecessary findings such as false positives and test code are automatically dismissed, with up to 1,000 vulnerabilities processed per run.

📚 Sources

• 🔗 news.microsoft.com: For KPMG Canada’s Christine Andrew, Copilot isn’t just a time saver—it…
• 🔗 aws.amazon.com: Amazon Timestream for InfluxDB is now available in the Mexico (Central…
• 🔗 blog.cloudflare.com: Sandboxing AI agents, 100x faster…
• 🔗 about.gitlab.com: Manage vulnerability noise at scale with auto-dismiss policies…
• 🔗 NVIDIA: Blowing Off Steam: How Power-Flexible AI Factories Can Stabilize the G…

Want to generate content for all your website’s products with a single click?

Visit our website — anishoai.com