Nov 1: Attackers exploit Cisco ASA VPN flaw to reboot firewalls
Today's 3 things that matter
- Attackers exploit Cisco ASA VPN flaw to reboot firewalls
Over 51,000 internet-exposed ASA SSL VPN devices exist; this zero-day disrupts VPN access for remote workers with no authentication required. Network ops must patch immediately or disable secondary authentication on tunnel groups as a workaround. - AI-native telco networks reach production at major carriers
Operators are standardizing on AI infrastructure plays; network teams must align telemetry stacks, revisit silicon/optical interconnect choices for AI workloads, and plan AI-aware APIs to avoid vendor lock-in and ensure cloud-network roadmap coherence over the next 12-18 months. - IETF draft carries BGP monitoring protocol over QUIC
Changes how you transport BGP monitoring data—QUIC's multiplexing reduces connection overhead and latency compared to TCP-only BMP, critical for large-scale route monitoring at hyperscale and carrier deployments.
Full stories below, grouped by topic.
AIOps & Network Observability
Forrester names Dynatrace a leader in AIOps platforms
Dynatrace · Oct 30, 2026 · Vendor release
What happened: Forrester named Dynatrace a Leader in The Forrester Wave: AIOps Platforms, Q4 2026, evaluating 10 providers on 24 criteria with highest scores in Current Offering category and 16 individual criteria. Dynatrace's hypermodal AI integrates causal, predictive, and generative AI for deterministic root-cause analysis.
Why it matters: Dynatrace's causal AI approach grounds AIOps in real-time topology knowledge, not just correlation, enabling NOCs to automate with confidence and reduce MTTR on complex hybrid infrastructure.
Forrester's recognition validates a key architectural difference: while most AIOps platforms correlate alerts probabilistically, Dynatrace determines cause-and-effect relationships through its Smartscape topology model. The platform's Grail data store and Dynatrace Intelligence umbrella (causal, predictive, generative, and agentic AI) create a unified context layer across metrics, logs, traces, events, and security data. Critically for NOC operations, this means agents and humans act on facts, not guesses. The MCP (Model Context Protocol) server is GA, enabling external AI assistants to query Dynatrace safely. With the October 2026 acquisition of Arize completed, AI and LLM observability capabilities are being integrated into the platform, addressing a key gap in monitoring data center AI workloads. For operations teams evaluating AIOps platforms, this report highlights the maturation of deterministic AI versus statistical correlation as the competitive frontier.
Read the original at dynatrace.com
Most telecom operators plan network digital twins for autonomy
Global Risk Community · Oct 27, 2026 · Industry news
What happened: TM Forum survey of 128 communications service providers found 72% plan to use network digital twins to support autonomous networks and agentic AI, with 13% already adopted. Leading trend is convergence of digital twins with agentic AI and closed-loop automation to validate autonomous actions against a model before execution.
Why it matters: Digital twins enable NOCs to validate agentic system changes before deployment by running them against a mathematical model of network behavior, transforming autonomous operations from risky to provably safe.
Gartner's 2026 Hype Cycle for Enterprise Networking identifies network digital twins as a transformational technology. The practice is shifting from mapping and visualization to predictive validation: before an autonomous agent makes a change, the digital twin answers "will this break reachability, violate segmentation policy, or cause a cascade?" Forward Networks' approach builds mathematically accurate, continuously updated models from device state and configuration data across on-premises and cloud infrastructure. Key players include telecom equipment makers (Nokia, Ericsson, Huawei, ZTE) and specialists like Forward Networks (network modeling and change validation) and Keysight (emulation and testing). For NOC operations, this maturation means agentic systems can operate at scale: instead of autonomous agents executing changes and waiting for alerts, they execute changes validated against a verified model of network behavior. Adoption intent is high among service providers facing pressure to reduce operational overhead while scaling AI workloads.
Read the original at globalriskcommunity.com
Network Security
Attackers exploit Cisco ASA VPN flaw to reboot firewalls
BleepingComputer · Oct 29, 2026 · Industry news
What happened: Cisco disclosed CVE-2026-76418, an unauthenticated denial-of-service vulnerability in ASA and FTD firewalls that allows remote attackers to force device reboot via crafted SSL VPN requests. Active exploitation confirmed in October 2026; CISA added to KEV catalog October 29.
Why it matters: Over 51,000 internet-exposed ASA SSL VPN devices exist; this zero-day disrupts VPN access for remote workers with no authentication required. Network ops must patch immediately or disable secondary authentication on tunnel groups as a workaround.
CVE-2026-76418 (CVSS 8.6, High) affects Cisco Secure Firewall ASA and Firepower Threat Defense (FTD) running software with webvpn configuration and SSL VPN enabled. The vulnerability stems from logic errors in secondary authentication processing on tunnel groups. An unauthenticated attacker sends a crafted request to the SSL VPN login path, causing the firewall to reload without successful VPN authentication. Cisco PSIRT became aware of active exploitation in October 2026. Cisco provided no workarounds; the recommended mitigation is to disable secondary authentication on DefaultWEBVPNGroup or upgrade to fixed releases. Shadowserver data shows 51,000+ exposed devices. Federal agencies were ordered to patch by November 1, 2026. This is the second exploited VPN crash vulnerability in ASA/FTD since August 2026.
Read the original at bleepingcomputer.com
GoAhead web server accepts revoked client certificates
The Hacker News · Oct 27, 2026 · Industry news
What happened: GoAhead 6.1.0 and earlier accept revoked client certificates on OpenSSL builds despite configured revocation lists (CVE-2026-115881, CVSS 9.1). A separate LibreSSL variant allows over-length certificate chains to bypass chain-length validation, bypassing mutual TLS authentication on affected appliances.
Why it matters: Network appliances using GoAhead with OpenSSL-backed mutual TLS (VPN gateways, API servers, management interfaces) may silently accept revoked client credentials. Audit and patch GoAhead to 6.1.1 immediately, or disable certificate revocation checks and validate manually.
CVE-2026-115881 affects GoAhead builds using OpenSSL backend with mutual TLS, a trusted CA bundle, issuer checking enabled, and a configured revocation list. The flaw silently accepts client certificates that are present on the revocation list, allowing revoked credentials to authenticate. An attacker holding a certificate signed by a trusted authority but listed as revoked can bypass authentication. CVE-2026-115884 (CVSS 7.4) on LibreSSL allows attackers to present certificate chains exceeding GoAhead's 10-certificate limit and complete handshakes as authenticated clients without trusted keys. OpenSSL, BoringSSL, AWS-LC, and LibreSSL 4.x reject such chains. Mitigation requires upgrading to 6.1.1, enabling issuer checking with a trusted CA bundle, or switching to LibreSSL 4.x. This affects network infrastructure relying on mTLS for machine-to-machine authentication.
Read the original at thehackernews.com
Telco & Cable AI
Operators move agentic AI into production network operations
TechSabado, Fierce Network, VoIP Review · Oct 29, 2026 · Industry news
What happened: Telecom operators are moving agentic AI from pilots into production network operations, with 65% of telecom operators reporting AI drives network automation and 59% identifying it as a leading use case with measurable ROI. Network automation has overtaken customer experience as the top AI deployment priority.
Why it matters: Network automation is now the primary telco AI use case with proven returns; NetDevOps and operations teams must accelerate agentic AI adoption for incident detection and root-cause analysis to remain competitive as operators shift from 71% partial automation to full autonomy targets.
Network automation was cited as an AI use case by 54% of telecommunications respondents in 2026 surveys, up 17 percentage points from 2024, with 65% of operators saying AI drives network automation and 59% identifying it as generating measurable ROI. Among operators who deployed agentic AI, 41% used it for network automation, with 52% expecting network automation agents to deliver significant returns within two years. Orange, Telenor and Vodafone in a recent webinar highlighted priorities: faster operations, improved accuracy in root-cause analysis, and automation supporting revenue, loyalty and trust. The same teams run networks after AI deployment without staffing increases, indicating true efficiency gains. ServiceNow reported 90% reduction in network incident volume after moving to Juniper Mist AI, with a 2028 target for full autonomy. Current state: 71% of operators have partial or conditional automation, 25% remain at basic or no automation levels, indicating significant deployment variability across the industry.
Read the original at techsabado.com
AI-native telco networks reach production at major carriers
TelecomTV · Oct 26, 2026 · Industry news
What happened: AI-native network operations moved into production reality across major carriers in late October 2026, with T-Mobile completing first AI-RAN calls and ServiceNow achieving 90% incident reduction using HPE Juniper Mist AI. Operators are now planning AI-aware network APIs and telco-trained models as production services.
Why it matters: Operators are standardizing on AI infrastructure plays; network teams must align telemetry stacks, revisit silicon/optical interconnect choices for AI workloads, and plan AI-aware APIs to avoid vendor lock-in and ensure cloud-network roadmap coherence over the next 12-18 months.
The week of October 17-23, 2026 saw AI operations and AI-RAN advances move into production across multiple operators. T-Mobile completed its first AI-RAN call on a live network in Bellevue; ServiceNow reported 90% reduction in network incidents using HPE Juniper Mist AI with a 2028 full-autonomy target. AT&T and GSMA leaders are now discussing telco-trained models, smart routing, and guardrails as first-class network capabilities, pointing toward production AI-enhanced network APIs and managed AI connectivity services. Engineering guidance from industry analysts calls for: unified telemetry and control stacks for AI; NTN (non-terrestrial network) and D2D assumptions in spectrum and device roadmaps; and joint review of AI workloads, optical transport, and silicon choices. Operators are being advised to expose AI-aware network APIs and connectivity products within 12 to 18 months to capture emerging revenue from enterprise AI deployments.
Read the original at telecomtv.com
Research, Standards & Industry
IETF draft carries BGP monitoring protocol over QUIC
IETF Datatracker · Oct 27, 2026 · Primary source
What happened: IETF GROW working group updated draft for BGP Monitoring Protocol over QUIC transport, enabling multiple simultaneous streams and improved efficiency for route monitoring. Version 01 replaces earlier draft with technical refinements for operational monitoring.
Why it matters: Changes how you transport BGP monitoring data—QUIC's multiplexing reduces connection overhead and latency compared to TCP-only BMP, critical for large-scale route monitoring at hyperscale and carrier deployments.
The GROW working group continues advancing BMP (BGP Monitoring Protocol) beyond its traditional TCP-only transport. Version 01 of draft-ietf-grow-bmp-over-quic (published 2026-10-27) leverages QUIC's native multi-stream capabilities to address operational limitations in monitoring large numbers of BGP peers. QUIC provides connection multiplexing, 0-RTT resumption, and better packet loss recovery than TCP, reducing overhead on monitoring systems collecting route views from hundreds or thousands of routers. This is particularly valuable for operators running centralized monitoring infrastructure or managing geographically distributed vantage points. The draft maintains full backward compatibility with existing BMP semantics while opening a path to deprecate TCP as monitoring scale increases. Key authors include Thomas Graf and Paolo Lucente, both active in operational telemetry standards.
Read the original at datatracker.ietf.org
OpenConfig merges vendor-neutral YANG model for ICMP
OpenConfig GitHub · Oct 25, 2026 · Primary source
What happened: OpenConfig repository merged new openconfig-icmp YANG module providing vendor-neutral configuration and operational state for ICMPv4 and ICMPv6 per network instance. Augments network-instance subtree with standardized ICMP parameters and counters.
Why it matters: Standardizes ICMP configuration management across multi-vendor networks—enables consistent, model-driven control of echo requests, time-exceeded handling, and error message behavior per network context without vendor-specific CLI variation.
OpenConfig contributors added a dedicated ICMP YANG model (openconfig-icmp) on 2026-10-25, addressing a gap in vendor-neutral network management. The module augments the network-instance hierarchy, allowing per-VRF or per-namespace ICMP tuning—critical for multi-tenant and segment-routing deployments where different services need different ICMP behavior. The model includes both configuration (IPv4 and IPv6 parameters) and operational counters (echo requests, unreachables, time-exceeded, parameter-problem). This follows OpenConfig's operationally-complete approach: focusing on the most commonly tuned parameters rather than exhaustive protocol coverage. Integration with gNMI/NETCONF makes it immediately usable in NetDevOps pipelines alongside existing OpenConfig models for interfaces and routing. Contributors reviewed for parity with Cisco, Juniper, and Arista implementations.
Read the original at github.com
AI Industry & Policy
Big Tech borrows record sums to fund AI buildouts
Digital Anthropology News Digest · Oct 31, 2026 · Analysis
What happened: Big Tech now borrows to build AI while Korean chipmakers book the profit; Alphabet sold $80.8 billion of bonds in nine months and spent more than it earned in cash; Amazon's trailing free cash flow swung to an outflow of $42.4 billion; AI-linked borrowers sold $368 billion of bonds this year, more than six times last year's amount. Moscow's investment envoy marketed energy projects and AI data centers to American partners, indicating active dialogue on building AI infrastructure.
Why it matters: AI infrastructure financing is shifting from equity to debt markets, and geopolitical competition for compute capacity and semiconductor supply is reshaping capital allocation and partner relationships—directly impacting infrastructure procurement strategy and risk exposure for enterprises.
The October 31 analysis reveals two structural shifts reshaping AI economics. First, Big Tech's debt accumulation for AI buildout is accelerating: bond issuance by AI-linked borrowers hit $368 billion in 2026—six times 2025's volume—indicating equity markets cannot absorb the capital requirements for data center expansion at current valuations. Alphabet and Amazon's cash flow dynamics are particularly revealing: despite generating substantial revenue, both are spending beyond operating cash flow on AI infrastructure, forcing debt financing and raising questions about returns on deployed capital. This mirrors historical infrastructure booms (canals, railroads, electrification, telecommunications) but with far higher volatility and speculative risk embedded in financial structures. Second, geopolitical competition for AI sovereignty and compute capacity is restructuring state-level investment and international negotiations. Russia's active outreach to the US on AI data center partnerships—alongside Moscow's concurrent engagement with Beijing—indicates AI infrastructure is becoming a recognized geopolitical asset and bargaining chip alongside energy. This creates new supply-chain and regulatory risks for enterprises: chip availability, compute capacity pricing, and foreign investment restrictions may become operational constraints rather than optional considerations.
Read the original at olegov.substack.com
Read this edition on the web · The week in network intelligence · Vendor Radar
Digital Plumber is AI-curated and AI-summarized, with no human review before publishing. Verify before acting on anything here. How it works.