Tech News: Latest Updates 2026-03-11
This report is a summary of technology news from 2026-03-11 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.
Microsoft Announces Quarterly Dividend
According to news.microsoft.com, Microsoft Corp. announced on Tuesday that its board of directors has approved a quarterly dividend of $0.91 per share.
| Details | Value |
|---|---|
| Dividend per share | $0.91 |
| Payment date | June 11, 2026 |
| Record date | May 21, 2026 |
| Ex-dividend date | May 21, 2026 |
This dividend will be paid on June 11, 2026 to shareholders of record as of May 21, 2026.
Microsoft (Nasdaq: MSFT) develops AI-powered platforms and tools. The company’s mission is to empower every person and every organization on the planet to achieve more.
Amazon Bedrock AgentCore Support for Stateful MCP
According to aws.amazon.com, the Amazon Bedrock AgentCore Runtime service now supports Stateful capabilities of the Model Context Protocol (MCP). This capability allows developers to build MCP servers that leverage elicitation, sampling, and progress notifications alongside existing support for resources, prompts, and tools.
In stateful MCP sessions, each user session runs in a dedicated microVM with isolated resources, and the server maintains session context across multiple interactions using the Mcp-Session-Id header. The elicitation capability enables server-initiated multi-turn conversations to gather information such as user preferences. The sampling capability allows servers to request AI-powered text generation from the client, for example for personalized recommendations. Progress notifications keep clients informed during long-running operations such as flight searches or booking processing.
| AWS Region |
|---|
| US East (N. Virginia) |
| US East (Ohio) |
| US West (Oregon) |
| Asia Pacific (Mumbai) |
| Asia Pacific (Seoul) |
| Asia Pacific (Singapore) |
| Asia Pacific (Sydney) |
| Asia Pacific (Tokyo) |
| Canada (Central) |
| Europe (Frankfurt) |
| Europe (Ireland) |
| Europe (London) |
| Europe (Paris) |
| Europe (Stockholm) |
These stateful MCP capabilities are available in fourteen AWS regions. Together, these features support complex, interactive agent workflows that go beyond simple request-response patterns.
Investigating Multi-Vector Attacks with Cloudflare Log Explorer
According to blog.cloudflare.com, the Cloudflare Log Explorer tool provides full coverage across Application Services and Cloudflare One products by integrating 14 new datasets. By correlating application-layer data from HTTP requests, network-layer logs related to DDoS and Firewall, and Zero Trust Access events, this tool helps security teams reduce Mean Time to Detect (MTTD) and identify complex multi-layered attacks. Logs serve as a flight recorder for the entire infrastructure, capturing every interaction and attack attempt before it reaches the server.
| Log Type | Scope | Use Case |
|---|---|---|
| HTTP Requests, Firewall Events, DNS | Zone-Scoped | Website traffic, security events, and edge network performance |
| Access Requests, Gateway, Audit | Account-Scoped | Internal security, Zero Trust, administrative changes |
| Magic IDS, Network Analytics | Account-Scoped | Detecting volumetric attacks and East-West lateral movement |
According to this report, Cloudflare Log Explorer enables detection of malicious activities at various stages of an attack. For example, during the initial reconnaissance phase, one can examine 401, 403, or 404 status codes from a single IP or requests to sensitive paths (such as /.env, /.git, /wp-admin). Magic IDS logs also have the capability to detect scans such as Nmap and SYN stealth scans at the TCP and UDP layers. During the attack investigation phase, fields such as WAFAttackScore and WAFSQLiAttackScore in fw_events logs indicate whether malicious payloads (such as SQL injection) have bypassed defenses. Managed Rules block many of these attacks by default.
This tool also enables investigation of identity breaches; by searching for ClientIP in access_requests, if a Decision: Allow status is observed for a sensitive internal application, it indicates a compromised user account. Regarding data exfiltration prevention, attackers sometimes use DNS tunneling to bypass firewalls, encoding sensitive data such as passwords or SSH keys in DNS queries, which can be detected by examining high-entropy subdomains.
Automating Detection Gap Analysis with GitLab Duo Agent Platform
According to about.gitlab.com, the Signals Engineering team at GitLab has automated the Detection Gap Analysis process using the GitLab Duo Agent Platform. A detection gap occurs when an attacker performs an action but detection systems fail to identify it. Manual review of these gaps is a time-consuming, inconsistent process that is prone to being deprioritized.
The GitLab Duo Agent Platform is a framework for building and deploying AI agents that can reason, take action, and integrate with GitLab resources such as issues and merge requests. The platform offers two practical paths: using out-of-the-box agents such as the Security Analyst Agent, or building custom agents. The Signals Engineering team has leveraged both paths and built their custom agent called the Detection Engineering Assistant.
The Security Analyst Agent, which comes built-in, can be invoked directly from an issue and by reviewing the incident content, map undetected tactics and techniques based on MITRE ATT&CK. This agent is suitable for quick starts without requiring configuration, but it is not aware of the team’s dedicated environment including SIEM, log sources, and domain-specific standards. For this reason, the GitLab team built the custom Detection Engineering Assistant agent, which operates like an expert team member through a precise and comprehensive system prompt.
Building a custom agent in the Duo Agent Platform involves specifying a name, description, and a system prompt. According to this report, a vague system prompt produces weak output, while a precise and comprehensive system prompt creates an agent that behaves according to the security team’s needs.
NVIDIA Jetson and Running Open AI Models at the Edge
According to NVIDIA, the NVIDIA Jetson platform has become a standard for running open AI models at the Edge. At the CES exhibition, the Cat AI Assistant ran on NVIDIA Jetson Thor; this assistant is installed in the Cat 306 CR excavator and uses NVIDIA Nemotron speech models for voice interaction and the Qwen3 4B model via vLLM for processing requests locally without requiring cloud connectivity.
Running OpenClaw on NVIDIA Jetson enables developers to build private, always-on AI assistants. According to NVIDIA, all Jetson developer kits support OpenClaw and provide the flexibility to switch between open models ranging from 2 billion to 30 billion parameters. The entry-level model starts with the Jetson Orin Nano 8GB.
| Project / Team | Platform | Details |
|---|---|---|
| Franka Robotics (FR3 Duo) | NVIDIA Jetson | Running the GR00T N1.6 model end-to-end on device |
| SONIC (GEAR Lab) | Jetson Orin | Trained with over 100 million motion-capture frames; policy loop at 50 Hz |
| SIGRobotics (UIUC) | Jetson Thor | Matcha-making robot with GR00T N1.5 model; first place at NVIDIA hackathon |
| NYU Center for Robotics and Embodied Intelligence | Jetson Thor + NVIDIA Blackwell | YOR robot for pick-and-place tasks |
| Andrés Marafioti (Hugging Face) | Jetson AGX Orin | Agentic AI system with model routing |
| Ajeet Singh Raina (Collabnix) | NVIDIA Jetson Thor | Personal assistant with OpenClaw and local inference |
This platform accelerates customer hardware design by combining compute and memory in a single system-on-module. Developer Ajeet Singh Raina from the Collabnix community has demonstrated how to build a personal assistant with complete data privacy by running OpenClaw on NVIDIA Jetson Thor, managing emails and calendar through a local gateway.
Summary
Artificial intelligence is the core driver of this week’s technology developments; from Amazon Bedrock AgentCore’s support for the stateful MCP protocol for building intelligent agents, to automating security analysis with GitLab Duo Agent Platform, and running open AI models on the NVIDIA Jetson platform at the edge. Meanwhile, Cloudflare has strengthened its multi-vector attack analysis capabilities by integrating 14 new datasets into Log Explorer, and Microsoft has announced a quarterly dividend of $0.91 per share.
Frequently Asked Questions
What new capabilities does Amazon Bedrock AgentCore Runtime offer for the MCP protocol?
This service now supports stateful MCP capabilities including elicitation, sampling, and progress notifications; each user session runs in a dedicated microVM with isolated resources, and the server maintains session context using the Mcp-Session-Id header.
How many new datasets has Cloudflare Log Explorer integrated and what is its use case?
This tool has integrated 14 new datasets and by correlating HTTP data, DDoS and Firewall logs, and Zero Trust Access events, it helps security teams reduce Mean Time to Detect and identify multi-layered attacks.
What range of AI models does the NVIDIA Jetson platform support?
All Jetson developer kits support OpenClaw and provide the flexibility to switch between open models ranging from 2 to 30 billion parameters; for example, the Cat AI Assistant ran on Jetson Thor with Nemotron and Qwen3 4B models locally without cloud connectivity.
📚 Sources
- 🔗 news.microsoft.com: Microsoft announces quarterly dividend…
- 🔗 aws.amazon.com: Amazon Bedrock AgentCore Runtime now supports stateful MCP server feat…
- 🔗 blog.cloudflare.com: Investigating multi-vector attacks in Log Explorer…
- 🔗 about.gitlab.com: Automating detection gap analysis with GitLab Duo Agent Platform…
- 🔗 NVIDIA: As Open Models Spark AI Boom, NVIDIA Jetson Brings It to Life at the E…
Want to generate content for all your website’s products with a single click?
