You're reading the Monday, October 26, 2026 edition. Today's briefing →
Toronto
Monday, October 26, 2026No. 134

Digital Plumber

Plumbing the information age

AI-curated intelligence for people who run networks. Daily coverage of AIOps, network automation, agentic operations, AI infrastructure, security and the vendors shaping them.

Today's 3 things that matter

Picked by the AI editor
  1. Security·Industry news

    HPE patches eight CVSS 10.0 flaws in Aruba wireless systems

    HPE patched 72 vulnerabilities in Aruba AOS-8 and AOS-10 wireless operating systems on October 22, 2026, including eight maximum-severity authentication bypass and remote code execution flaws rated CVSS 10.0 exploitable without credentials through management interfaces and PAPI protocol.

    Why it matters These flaws affect enterprise wireless infrastructure trusted by 90% of Fortune 500 companies; attackers can gain administrative control of mobility conductors, controllers, gateways, and access points without authentication through web management interfaces and PAPI.

  2. Routing·Primary source

    RIPE NCC adds BGPsec router key certificates to RPKI API

    RIPE NCC has introduced support for creating and managing BGPsec Router Key Certificates through the RPKI API, enabling AS holders to authorize router keys for signing BGP updates.

    Why it matters Operators can now use the RPKI API to manage BGPsec keys programmatically, advancing cryptographic BGP security beyond ROA-based origin validation toward path-level authentication across the routing system.

  3. Agentic AI·Industry news

    Only 18% of enterprises let network agents fix things

    New research shows most enterprises still use network agents to advise rather than act.

    Why it matters Adoption data reveals that most organizations test agents as advisers first, not autonomous actors—network teams planning remediation automation must account for governance and approval workflows as operational baseline.

Today's briefing

What happened, and why it matters

14 stories · 10 topics · Updated 12:50 PM ET

AutoCon 6 talk dissects automating from spreadsheet sources of truth

NIL (part of Conscia) · Oct 19, 2026 · Industry news

AutoCon 6 session abstract examining failure modes when automating network changes directly from manually-maintained spreadsheets. Covers data ownership, validation, governance, and scalability pitfalls with practical migration strategies to proper source-of-truth platforms like NetBox.

Why it matters Warns practitioners that source-of-truth data quality matters more than tooling—poor data governance embedded in automation exposes hidden process weaknesses and operational risks at scale.

Toni Sarađan's AutoCon 6 presentation (November 19, 2026) explores a scenario many network engineers recognize: Excel spreadsheets managing critical network information that become the foundation for automation projects. What appears initially successful quickly reveals broader challenges: data ownership gaps, lack of validation logic, governance failures, ad-hoc change control, and scaling brittleness. The session highlights practical lessons from automating against manually-maintained sources: when spreadsheets become the source of truth for network automation, operational risk increases dramatically because changes lack validation gates, audit trails are incomplete, and data consistency cannot be enforced. The presentation demonstrates real-world examples and live demonstrations of network changes automated directly from spreadsheet-based data sources, then contrasts this anti-pattern with proper migration strategies toward dedicated platforms. Key takeaway: the quality of source-of-truth data—validation logic, change control, and governance—matters far more than which tool you choose. Attendees gain practical insights into common failure modes and techniques for reducing risk through validation and governance, positioning NetBox (or Nautobot) adoption as a critical prerequisite for safe, scalable automation rather than a tool decision.

Read the original at nil.com ↗

OpenTelemetry interoperability becomes the battleground for AI observability

The New Stack · Oct 26, 2026 · Industry news

OpenTelemetry and interoperability standards are becoming critical for enterprise observability and AIOps. Teams standardizing telemetry through OTel can govern AI agent behavior and measure cost per business outcome without vendor lock-in.

Why it matters For AIOps practitioners: OTel interoperability reduces vendor lock-in and enables multi-backend telemetry processing, essential for building observable, auditable AI-driven operations without tool sprawl.

The article emphasizes that the competitive advantage in AI observability will be determined by how well platforms support interoperability and open standards rather than proprietary capabilities alone. OpenTelemetry and Prometheus integration is advancing the ability to standardize how telemetry is captured, structured, and shared across applications, clouds, and vendors. This matters for NOC operations because fragmented observability data prevents AI models from reasoning effectively across infrastructure layers. When telemetry is unified via OTel, AIOps systems can perform better event correlation, root-cause analysis, and automated remediation. The article notes that interoperability also enables governance frameworks—organizations can inspect local telemetry for OTel compliance before deployment and validate whether instrumentation follows conventions. For network operations teams evaluating AIOps platforms, this means selecting tools with native OTel support and multi-backend export capabilities protects against future platform migrations and ensures data portability. The piece is published on The New Stack, a trusted source for infrastructure and observability analysis, and positions interoperability as the decisive factor in AIOps maturity—not raw AI model quality alone.

Read the original at thenewstack.io ↗

Ultra Ethernet aims to match InfiniBand as Broadcom silicon nears

Spheron · Oct 19, 2026 · Analysis

The Ultra Ethernet Consortium, formed in 2023 with AMD, Intel, HPE, Broadcom, Cisco, Meta, and Microsoft as founding members, aims for an open Ethernet RDMA fabric that matches InfiniBand latency without proprietary components. UEC 1.0 (published 2024) defines credit-based reliability in NIC firmware, eliminating head-of-line blocking. Broadcom's Tomahawk-UEC silicon is expected in 2026–2027.

Why it matters UEC 1.0 eliminates head-of-line blocking that makes RoCEv2 hard to operate at scale. This October 19 update provides operators a current decision framework: UEC hardware is still pre-production, making Spectrum-X the best production Ethernet option today if you need performance beyond vanilla RoCE.

The Ultra Ethernet Consortium, established in 2023 with AMD, Intel, HPE, Broadcom, Cisco, Meta, and Microsoft, targets an open Ethernet RDMA fabric for AI that matches InfiniBand latency without proprietary components. UEC 1.0 (published 2024) defines a transport protocol (Ultra Ethernet Transport) that avoids PFC entirely, using credit-based reliability implemented in NIC firmware rather than relying on switch-level pause frames, eliminating the head-of-line blocking problem that makes RoCEv2 difficult to operate at scale. UEC 1.0 also specifies hardware-offloaded collectives at the NIC level, not the switch level. Broadcom's Tomahawk-UEC silicon targeting UEC 1.0 is expected in 2026–2027. Until products ship, Spectrum-X is the best Ethernet option if you need better-than-vanilla-RoCE performance today. The article was updated October 19, 2026, providing a current practitioner snapshot of the InfiniBand-vs-Ethernet-vs-RoCE landscape as hardware transitions from Hopper (H100/H200) to Blackwell (B200) clusters.

Read the original at spheron.network ↗

RIPE NCC adds BGPsec router key certificates to RPKI API

RIPE NCC · Oct 23, 2026 · Primary source

RIPE NCC has introduced support for creating and managing BGPsec Router Key Certificates through the RPKI API, enabling AS holders to authorize router keys for signing BGP updates. This extends RPKI cryptographic protection beyond route origin validation to include path validation.

Why it matters Operators can now use the RPKI API to manage BGPsec keys programmatically, advancing cryptographic BGP security beyond ROA-based origin validation toward path-level authentication across the routing system.

The RIPE NCC Member Update for October 2026 announced a significant expansion of RPKI management capabilities: BGPsec Router Key Certificates are now supported through the RPKI API. This allows AS holders to create, manage, and revoke router keys used for signing BGP updates, adding a cryptographic layer of protection to BGP path validation. BGPsec Router Key Certificates complement existing Route Origin Authorization (ROA) capabilities by enabling end-to-end cryptographic verification of AS paths, not just route origins. This is a progression toward RFC 8205 BGPsec deployment, where routers can cryptographically sign updates and validate signatures from neighboring ASes. For operators managing large numbers of routers or multi-vendor environments, API-based management of BGPsec keys reduces manual provisioning overhead and improves auditability. However, adoption remains limited: BGPsec deployment is still not widespread on the global Internet, as it requires coordinated upgrades across ASes and carries operational complexity around key rotation and management.

Read the original at ripe.net ↗

Cloudflare Tunnel adds URL normalization for ingress rule matching

Cloudflare · Oct 22, 2026 · Primary source

Cloudflare Tunnel now supports URL normalization before matching ingress rules, starting with cloudflared 2026.10.0. The feature normalizes URLs for Access validation checks while keeping the path sent to origins unchanged; it will default to on from January 25, 2027.

Why it matters Network and security teams using Cloudflare Tunnel path-based routing must test their configurations before January 2027, as default URL normalization may change which ingress rule matches requests and affect traffic routing behavior.

Starting from cloudflared version 2026.10.0, Cloudflare Tunnel can normalize URLs before matching them against ingress rules. This feature improves security for tunnels with rules that require Cloudflare Access validation by standardizing URL format and reducing ambiguity in path-based rule matching. However, the URL path sent to the origin application remains unmodified, preserving backend application behavior. The feature is currently opt-in (off by default), but will become the default in new cloudflared releases starting January 25, 2027. This is a significant timeline for operators to test: any teams using path-based routing rules should validate their configurations before the default flip, as URL normalization can change which ingress rule matches a request—potentially redirecting traffic to different backends or requiring different Access policies. Configurations without rules requiring Access validation are unaffected. This change is part of Cloudflare's broader Tunnel hardening work and reflects a shift toward stricter URL parsing for security-sensitive routing decisions.

Read the original at developers.cloudflare.com ↗

Eight eMudhra intermediate CA keys recovered from trivially weak entropy

The Hacker News · Oct 20, 2026 · Industry news

Eight eMudhra intermediate CA private keys were compromised with catastrophically weak entropy—3 to 8 bits instead of 384 bits required for NIST P-384. Chrome and Firefox blocked the eight certificates within hours of notification on October 17.

Why it matters Six keys could issue TLS certificates and two code-signing certificates; attackers could forge certificates for any website or service until revocation blocks distributed, enabling man-in-the-middle attacks on encrypted communications.

The Indian Certificate Authority eMudhra issued eight TLS certificates bearing catastrophically weak private keys, ranging from just 3 to 8 bits of entropy instead of the standard 384 bits required for NIST P-384 elliptic curve cryptography. On October 17, 2026, a security researcher reported that the private keys of eight intermediate CA certificates could be recovered due to insufficient randomness in key generation; within two hours, Google and Mozilla pushed blocks to Chrome and Firefox. The incident escalated from tactical intermediate blocks to global root cert distrust across Mozilla, the Google Chrome Root Store, Apple, and Microsoft. On October 22, 2026, eMudhra announced its decision to exit the global public SSL/TLS certificate business entirely. For network and infrastructure teams, this compromises the trust model for any TLS connection anchored to these certificates. Organizations should urgently audit certificate chains used for internal services, VPN termination, and API endpoints, verify whether any systems trust the affected eMudhra root or intermediates, and update OS trust stores and certificate bundles. Chrome is removing emSign ECC Root CA – G3 from its root store on October 27, 2026, which will break legitimate certificates chained to it.

Read the original at thehackernews.com ↗

HPE patches eight CVSS 10.0 flaws in Aruba wireless systems

BleepingComputer · Oct 23, 2026 · Industry news

HPE patched 72 vulnerabilities in Aruba AOS-8 and AOS-10 wireless operating systems on October 22, 2026, including eight maximum-severity authentication bypass and remote code execution flaws rated CVSS 10.0 exploitable without credentials through management interfaces and PAPI protocol.

Why it matters These flaws affect enterprise wireless infrastructure trusted by 90% of Fortune 500 companies; attackers can gain administrative control of mobility conductors, controllers, gateways, and access points without authentication through web management interfaces and PAPI.

HPE Networking patched 72 vulnerabilities in wireless operating systems AOS-8 and AOS-10 (ArubaOS) on October 22, 2026 via advisory HPESBNW05162. Eight flaws are rated CVSS 10.0: CVE-2026-79844 (AOS-10 authentication bypass leading to RCE), CVE-2026-79845 (AOS-10 authentication bypass in PAPI protocol), CVE-2026-79847 and CVE-2026-79851 (unauthenticated buffer overflows leading to RCE in AOS-10), CVE-2026-79846 (unauthenticated command injection in AOS-8), CVE-2026-79848 and CVE-2026-79849 (unauthenticated SQL injection allowing RCE in AOS-8), and CVE-2026-79850 (authentication bypass leading to RCE in AOS-8). HPE urges customers to patch quickly and restrict management interface access to dedicated network segments; no public exploit code or in-the-wild exploitation has been observed. This is a critical patch day for any organization running Aruba wireless controllers and access points. Prioritize patching mobility conductors, controllers, and gateways that expose management interfaces to the network, and restrict management traffic to dedicated administrative VLANs or IP ranges with firewall rules.

Read the original at bleepingcomputer.com ↗

Kalepa names Garrett Schlesinger chief technology officer

Business Wire · Oct 22, 2026 · Vendor release

Kalepa appointed Garrett Schlesinger as Chief Technology Officer to lead applied AI, research, engineering, data, and infrastructure as the AI insurance platform expands globally. Schlesinger spent 12 years at Affirm building real-time underwriting systems and scaling infrastructure through hypergrowth.

Why it matters Brings proven production-scale AI infrastructure and team-scaling experience from fintech to lead enterprise agentic systems development; relevant for practitioners building automated workflow platforms.

Schlesinger joins Kalepa after tripling revenue and expanding product lines. At Affirm, he led engineering through every stage from seed to $20 billion public company valuation, overseeing real-time underwriting infrastructure, the Affirm Card launch, and system re-architecture supporting $50 billion+ annual transaction volume. His background in building scalable financial risk systems directly applies to Kalepa's AI-driven insurance underwriting, pricing, and portfolio management workflows. He joins alongside recent senior hires including CRO David Meister and Head of Product Audrey Howell as Kalepa builds executive depth.

Read the original at finance.yahoo.com ↗

Only 18% of enterprises let network agents fix things

Cybersecurity Institute · Oct 25, 2026 · Industry news

New research shows most enterprises still use network agents to advise rather than act. Only 18% let agents carry out fixes; 13.2% have agents acting on monitoring alerts, revealing adoption constraints for NetOps teams.

Why it matters Adoption data reveals that most organizations test agents as advisers first, not autonomous actors—network teams planning remediation automation must account for governance and approval workflows as operational baseline.

This week's Agentic NetOps coverage reveals adoption patterns: HPE Mist reduced network incidents by 90% via ServiceNow integration, Extreme Networks published first-month adoption metrics for Agent ONE, LogicMonitor enabled Edwin AI to operate as ServiceNow-callable agent, and Selector released root-cause agent with human-approval-only execution. Critical finding: 51% of enterprises in Cisco's Omdia survey run agentic AI in production, but only 18% permit agents to execute fixes; EMA found just 13.2% have agents acting on monitoring alerts, with production deployments limited to single sites. Running agents has surfaced cost concerns: Splunk found 57% of organizations pay more to monitor infrastructure than the infrastructure itself costs, with Bank of America raising price targets on Datadog and Dynatrace on expectations that AI usage will lift observability spend.

Read the original at cybersecurityinstitute.com ↗

AT&T and T-Mobile keep humans in the loop on AI

Reuters · Oct 23, 2026 · Industry news

AT&T and T-Mobile are investing in AI governance controls including encryption, multi-factor authentication, and human oversight as AI integration ramps up, though fully autonomous networks remain years away. AT&T has automated trouble-ticket analysis and repairs but emphasizes human intervention is still required for complex network problems.

Why it matters Network ops teams should expect staged automation rollouts with governance guardrails rather than full autonomy; full network autonomy remains years away, requiring human judgment for complex issues.

AT&T and T-Mobile are taking measured approaches to AI integration rather than rushing toward full autonomy. AT&T has deployed governance controls including encryption, multi-factor authentication, and human oversight of data access. The company has automated specific operational tasks—analyzing trouble tickets and implementing repairs—but CTO Jeremy Legg emphasized that fully autonomous networks remain years away because human judgment is still essential for complex problems. AT&T sees industry-wide security issues that individual operators cannot solve alone, raising questions about open-source security standards. T-Mobile similarly focuses on safeguarding customer data as AI agents automate workflows with less human intervention. Both operators view this as a multi-year journey requiring guardrails at each stage, not a sprint to full autonomy.

Read the original at whtc.com ↗

Lumen launches AI service assurance and dual path wavelengths

Simply Wall St · Oct 24, 2026 · Industry news

Lumen Technologies launched AI-powered Service Assurance tools and Dual Path wavelength service to address enterprise pain points around uptime, data control, and operational complexity across distributed AI environments. Snowflake partnership ties Lumen's programmable network to the AI Data Cloud.

Why it matters Lumen's shift to AI-ready programmable networking creates new demand signals for enterprise network upgrades supporting AI workloads, establishing a platform model that differs fundamentally from legacy circuit sales.

Lumen is repositioning from legacy telecommunications carrier into an AI infrastructure platform. The new AI-powered Service Assurance tools provide real-time network monitoring and predictive capabilities, while Dual Path wavelength ensures redundancy for critical enterprise workloads. The Snowflake partnership integrates Lumen's programmable network directly with cloud data platforms, enabling enterprises to move data from anywhere to anywhere in real time—critical for AI model training and inferencing at scale. This addresses a key market signal: enterprises increasingly realize that AI workloads require fundamentally different network characteristics than traditional applications. Lumen's message is that legacy point-to-point circuits cannot support AI; enterprises need intelligent, software-defined, low-latency connectivity. The company has exceeded 2,000 Network-as-a-Service customers and expects significant growth as enterprises operationalize AI.

Read the original at simplywall.st ↗

Anthropic previews Claude Duet as a proactive monitoring agent

CellCog AI · Oct 23, 2026 · Analysis

Anthropic opened a research preview of Claude Duet on October 21, 2026, a standalone app where Claude works as a proactive agent instead of waiting for you to open a chat. It monitors connected apps, sends daily briefings, and can handle tasks autonomously with user approval.

Why it matters Duet adds persistent agent capabilities to Claude's 6,000+ app connectors—meaning infrastructure teams running on Claude can offload task automation without polling or webhooks.

Duet is built around Opus 5.5 and available via web or standalone iOS app, limited to US Pro and Max subscribers on a waitlist. It delivers daily briefings and handles tasks from scheduling meetings to building financial forecasts and web apps, integrating with 6,000+ connectors including Gmail, Slack, Stripe Link, and Shopify Shop Pay. Anthropic implemented a new classifier and system prompt to prevent sensitive information leakage; Duet never directly sees passwords or sign-in tokens, and health or financial data only passes through with explicit user permission. Duet uses existing Claude plan limits, with background task checks counting toward them. The preview is gated initially to US subscribers, limiting early adoption scope.

Read the original at cellcog.ai ↗

683 AI rules now in force across 109 countries

MEAN CEO · Oct 24, 2026 · Analysis

The regulations.ai tracker counts 1,176 AI instruments across 109 countries, with 683 (58%) already in force. Enterprise buyers now send AI questionnaires before signing, and startups that answer in one day beat rivals needing three weeks and a lawyer.

Why it matters Compliance now works as a sales weapon in enterprise procurement, forcing vendor teams to integrate governance readiness into product delivery, not legal afterthought.

Most AI rules are already live and apply to your business today. The global regulatory landscape has shifted from draft guidance to enforcement: prohibited practices can cost up to €35 million or 7% of global turnover, with SMEs and startups paying lower amounts. The critical insight for enterprise teams: enterprise procurement will enforce AI rules before regulators do, with buyers' questionnaires already including AI Act sections. This creates a competitive gap where startups and vendors with documented compliance posture close deals faster. The 683 active instruments span multiple jurisdictions with divergent requirements, making compliance infrastructure a business enabler rather than pure cost center.

Read the original at blog.mean.ceo ↗

DeepSeek raises $15 billion in largest China AI round

bestproductivityresources.com · Oct 23, 2026 · Industry news

DeepSeek raised roughly $15B (100 billion yuan) including backing from CATL and Tencent, two days after the US ordered federal agencies to remove DeepSeek's technology. OpenAI is near a $30B round at $1.25T valuation, with Coatue and UAE-based investors leading the pre-IPO round.

Why it matters Geopolitical capital allocation is now actively shaping competitive AI landscape: US restrictions on Chinese AI accelerate capital concentration in rival centers while Western model providers consolidate funding, directly affecting enterprise procurement options and supply-chain sovereignty for AI infrastructure.

The simultaneous funding announcements expose deepening geopolitical fractures in AI development capital. DeepSeek's $15B round, reportedly including CATL and Tencent, arrived two days after the US ordered federal agencies to remove DeepSeek's technology—illustrating how export controls paradoxically trigger capital mobilization in sanctioned regions. Meanwhile, OpenAI's $30B pre-IPO round at $1.25T valuation, with Financial Times reporting OpenAI lowered its target from $1.4T after investor pushback, shows Western AI consolidation and potential valuation ceiling emerging. For enterprise teams, this bifurcation means: supply-chain risk increases as Chinese and Western AI infrastructure diverge further; model selection becomes a geopolitical choice (data residency, sanctions compliance); and long-term vendor stability depends on capital access in increasingly regionalized markets.

Read the original at bestproductivityresources.com ↗
Nothing in today's briefing matches that.

Listening

Podcasts and talks
  • Packet Pushers

    HN847: Building Trust In The Agentic NOC

    Joby Rudolph and Surya Nimmagadda from Selector.ai explain how AI agents work and how to design agents that build engineer trust in agentic AI for network operations, addressing widespread skepticism about non-deterministic AI behavior in production networks.

  • Packet Pushers

    TNO078: Designing Networks for Operations

    Bruce Davie and Larry Peterson discuss their new Computer Networks textbook edition and reality-based network operations, exploring the contrast between clean-slate tooling ideals and pragmatic realities of managing heterogeneous vendor environments with automation and verification.

  • Packet Pushers

    NAN134: Building Trust in AI-Driven Networking

    Eric Chou interviews Bob Friday (HPE Chief AI Officer) and Jose Tellado (VP Engineering) on agentic AI workflows and AI-native automation, discussing Large Experience Models, digital experience twins, and what agentic AI means for self-driving network operations.

Vendor Radar

Last 7 days · arrows compare with the 7 before

What changed this week

Last 7 days vs the 7 before

Biggest moves

Trending topics

agent · observability · OpenTelemetry · MCP · automation · RAG · Kubernetes · Agentic AI · AIOps · NetDevOps · SRE · inference

Get the daily briefing

Today's 3 things that matter and every story with why it matters, in your inbox each morning. Free, and you can unsubscribe at any time. Prefer a reader? Follow the RSS feed.