InfrastructureIssue 097 records
Utility Management
Tonight centers on regulatory packaging of water cyber and electric AI governance: the NARUC Annual Meeting draft PDF for 8–11 Nov includes WC-1 (WRRO minimums, CISA restaffing, FOIA-protected sharing) and EX-1 (EPRI SAFERai.power AI risk frameworks), while EPA’s advisory on AI providers requesting SCADA logs underscores OT data exposure. On operations, AEML states agentic outage AI must stay recommendation-only; AiDetect opens human-gated PACP coding to any U.S./Canada inspection software; Aalto’s peer-reviewed sewer ML+tracer rankings are only weakly concordant; Bluefield forecasts AI-native digital water and metering analytics growth; Blacktown’s vendor case shows stormwater CCTV AI informing CIP. US independent gas and transactional billing AI remain thin.
- Evidence records
- 7
- Cross-source patterns
- 3
- Evidence classes
- 2 independent reporting1 public-sector association guidance1 independent research1 academic research1 standards or public-body guidance1 vendor claim
- Outcomes
- 4 emerging2 cautionary1 mixed
- Source freshness
- 5 new this fortnight2 older, newly relevant
- Research completed
- 2026-11-07
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Utility AI touching OT or coded asset records should stay human-gated
AEML publicly restricts agentic outage orchestration to recommendations with authorised approval; AiDetect keeps NASSCO-certified and agency reviewers as the final say on PACP codes; EPA warns against unconstrained AI-provider access to SCADA logs. Across electric distribution, wastewater CCTV coding and water OT data, the operating pattern is human gates before actuation or system-of-record writes.
Operating questionWhere would an AI recommendation write to SCADA, OMS, or the asset register without a named human approver?
Supporting evidenceAdani Electricity Mumbai Limited (AEML) / Power LineUnderground Construction / ITpipesU.S. Environmental Protection Agency Office of Water (Emergency Response and Cybersecurity)
NARUC’s Nov meeting packages water cyber minimums with electric AI risk frameworks
The NARUC draft WC-1 urges a WRRO and restored CISA water assessments while EX-1 backs EPRI SAFERai.power for electric AI risk review; EPA’s advisory on AI requests for SCADA logs underscores why FOIA-protected sharing and OT data controls matter in the same week. Together they mark a regulatory packaging of cyber hygiene and AI governance for utilities.
Operating questionBefore 11 Nov, which OT remote-access and AI use-case gaps would a WRRO or SAFERai-style review expose first?
Supporting evidenceNational Association of Regulatory Utility Commissioners (NARUC)U.S. Environmental Protection Agency Office of Water (Emergency Response and Cybersecurity)
CCTV/condition AI for buried networks still needs corroboration before capital lock
Aalto’s peer-reviewed dual ranking of ML failure probability and tracer infiltration was only weakly/uncertainly concordant; AiDetect opens AI PACP coding with mandatory human QA; Blacktown’s vendor case uses AI CCTV features to inform CIP but without published accuracy trials. Capital use of inspection AI should retain independent evidence and sampled human review.
Operating questionWhat independent field or tracer evidence and QA sample will corroborate AI condition ranks before CIP approval?
Supporting evidenceAalto University / Journal of Environmental ManagementUnderground Construction / ITpipesBlacktown City Council / Vapar
Evidence ledger
Full record · every source keeps its link and limitations
NARUC 2026 Annual Meeting draft resolutions: WC-1 water cybersecurity WRRO and EX-1 EPRI SAFERai.power AI governance
Draft resolutions for the 8-11 Nov 2026 NARUC Annual Meeting include WC-1, urging Congress to reauthorize and restaff CISA (citing roughly one-third workforce loss since 2025 and September 2026 cuts to six free cyber assessments for water utilities), strengthen WaterISAC sharing with FOIA protections, and authorize EPA to create an independent non-governmental Water Risk and Resilience Organization (WRRO) for minimum drinking water and wastewater cybersecurity requirements. The draft cites attacks on water utilities in at least 12 states in July-August 2026 and notes only about 11% of community water systems fall under commission jurisdiction. EX-1 separately commends EPRI SAFERai.power for an operational risk framework and open-source tools assessing electric-utility AI by context, consequence, autonomy and risk, and supports state regulator engagement. Adoption date blanks remain open.
Why it matters, evidence and limitations
- Why it matters
- U.S. SLED water cyber posture and electric AI governance: a WRRO would push minimum requirements under EPA oversight, while SAFERai.power gives commissions shared language for AI scrutiny. Why now: vote window opens 8 Nov 2026; this PDF is the primary draft previously only summarized in press.
- Evidence and measured results
- Opened the official NARUC pubs PDF. WC-1 and EX-1 appear in full; TOC lists WC-1 dated 10/23/2026 (sponsor Commissioner Bange); NARUC list page marks a 3 Nov 2026 version. No final board adoption dates filled in.
- Limitations and uncertainty
- Draft only; Congress must still act for WRRO/CISA changes. SAFERai tools are under development. Incident and staffing claims are as stated in the draft.
AEML: integrating IT/OT, ML and agentic AI for distribution—recommendations only, humans approve execution
Power Line's 6 Nov 2026 feature on AEML describes SCADA, GIS, OMS and CRM in place, a data lake for KPIs, and ML for asset health, load forecasting and consumer segmentation under human oversight. Next focus is agentic AI correlating smart meters, SCADA, OMS, GIS and customer platforms to recommend outage restoration, with authorised personnel approving execution. The article warns uncontrolled agents could initiate actions with serious operational consequences, and cites cybersecurity, access controls, segmentation, data-protection law and skills as constraints. Stated 99.99% supply reliability is overall utility performance, not attributed to agentic AI, which remains prospective.
Why it matters, evidence and limitations
- Why it matters
- Independent scrutiny/control pattern for U.S. municipal and co-op distribution: agentic AMI/SCADA/OMS orchestration must stay recommendation-only. Transferability limited by Indian context; control posture aligns with NARUC SAFERai themes.
- Evidence and measured results
- Opened Power Line article dated 06 Nov 2026 10:57 IST, based on inputs from Rajesh Lad, Enterprise Architect, AEML. No third-party audit of agentic pilots.
- Limitations and uncertainty
- Utility-sourced via trade press; agentic orchestration proposed, not measured. India-specific regulation. Reliability stats are not an AI outcome.
Bluefield: European digital water spend to more than double by 2036, led by metering, analytics and AI-native tools
Water Magazine's 6 Nov 2026 summary of Bluefield's Europe Digital Water Landscape 2026-2036 projects annual digital water spend from US$12.6B to US$27.9B by 2036 (8.25% CAGR; ~US$211.8B cumulative). Metering and customer management grows at 7.8% CAGR, with meter data management and customer engagement at 10.1% and 14.5% CAGR as AMI expands. AI-native digital water solutions are forecast at 17.8% average annual growth—more than twice the overall market—building on existing monitoring. Conventional CCTV still accounts for 39% of spending across key wastewater network monitoring strategies. Information management is the fastest-growing category (11.3% CAGR) as cybersecurity needs rise with connected meters and sensors.
Why it matters, evidence and limitations
- Why it matters
- Fills metering/billing analytics gap with independent market evidence that AMI-adjacent MDM, customer platforms and AI-native analytics are where spend accelerates. U.S. municipals can use the maturity structure even though figures are European.
- Evidence and measured results
- Opened Water Magazine article dated 6 November 2026 quoting Bluefield analyst Maria Cardenal. Underlying Bluefield report paywalled; figures as reported in the summary.
- Limitations and uncertainty
- Forecast not measured outcomes; Europe-specific funding. Secondary summary of proprietary report. CCTV share shows AI inspection has not displaced conventional methods.
Integrating ML failure probabilities and tracer-based infiltration for sewer asset-management screening
A 1 Nov 2026 Journal of Environmental Management article (Vol. 419, 131130) combines IPCW-weighted ML failure probabilities from CCTV with independent dry-weather tracer infiltration estimates (EC and COD) across 21 operational zones. Zone rankings from ML structural failure probability and tracer infiltration severity showed positive but statistically uncertain agreement (Kendall tau-b 0.254-0.305; p 0.108-0.056; concordance index 0.628-0.655). Authors present dual-evidence screening as practical when CCTV coverage is incomplete, not as a validated pipe-level predictor. Full PDF returned 403 during this run; findings are from the opened abstract/record.
Why it matters, evidence and limitations
- Why it matters
- Methods lesson for U.S. municipal wastewater AI condition models: require independent corroboration and report uncertainty—do not treat ML rankings alone as capital truth.
- Evidence and measured results
- Opened Aalto portal record with abstract, journal, publication date 1 Nov 2026, and MoE A1 refereed type. DOI 10.1016/j.jenvman.2026.131130; PDF fetch blocked (403).
- Limitations and uncertainty
- Abstract-level inspection; PDF 403. Finnish network; uncertain statistical agreement means mixed. Not a product evaluation.
ITpipes opens AiDetect AI sewer defect coding to agencies and contractors using any inspection software
Underground Construction reported on 2 Nov 2026 that ITpipes opened AiDetect to wastewater agencies and contractors not using ITpipes software. Users upload a NASSCO exchange database with video; AiDetect assigns PACP code, distance and severity; NASSCO-certified reviewers check low-confidence items; agencies can verify against video before import. Model described as trained on 15 years of North American inspection data. CEO emphasises AI first pass with agency/contractor final say. Available now for mainline inspections in the U.S. and Canada. No measured coder-day savings or accuracy trial statistics appear in this article.
Why it matters, evidence and limitations
- Why it matters
- Addresses U.S. municipal wastewater workforce and backlog pressure with a human-gated AI coding path compatible with existing PACP workflows.
- Evidence and measured results
- Opened Underground Construction article dated 11/2/2026. Vendor claims via trade press; no independent accuracy study in-article.
- Limitations and uncertainty
- No published precision/recall; vendor-trained model; mainline only U.S./Canada. Trade press may reflect vendor messaging.
EPA advisory: AI service providers requesting water/wastewater SCADA and operational logs via public records
EPA’s 5 June 2026 advisory warns water and wastewater systems about disclosing sensitive operational information after multiple public-records requests from AI service providers, as reported by WaterISAC and AWWA. In at least one instance, an AI provider requested all SCADA logs for 2026 for treatment, distribution and wastewater systems, plus daily/hourly historical logs for flows, water-quality baselines, reservoir/tank levels and energy or equipment run-times. EPA concurs that aggregating such operational data across utilities could reveal patterns and vulnerabilities usable to disrupt service. Utilities are urged to make risk-informed decisions consistent with state information-protection laws. Why now: NARUC’s 8-11 Nov WC-1 draft seeks FOIA protections for cyber-incident sharing and WRRO minimums amid the same OT data-exposure risk surface.
Why it matters, evidence and limitations
- Why it matters
- Direct OT/AI security control for U.S. SLED water utilities: treat AI vendor data requests for SCADA/process historian extracts as high-risk even when framed as research or model training. Newly relevant alongside this week’s NARUC WRRO/CISA draft.
- Evidence and measured results
- Opened the EPA GovDelivery bulletin text dated 06/05/2026 09:50 AM EDT from Office of Water Emergency Response and Cybersecurity.
- Limitations and uncertainty
- Advisory dated June 2026 (older than last run) included for why-now linkage to NARUC WC-1 vote week. Does not name the requesting provider or quantify how many utilities received requests.
Blacktown City Council uses VAPAR AI on ~150 km of stormwater CCTV to prioritise capital works
Vapar’s case study states that since 2022 Blacktown City Council (largest NSW LGA) uploaded about 150,000 m (~150 km) of stormwater CCTV into the VAPAR platform, identifying over 20,000 pipe features and using risk-based models to prioritise rehabilitation versus maintenance. Insights informed capital works under the 2024–2025 Operational Plan, including allocation against a cited $7.7 million drainage improvements budget. Total Drain Group is named as systematically uploading footage. No independent audit of defect accuracy or quantified failure reduction is presented; the page’s Article JSON-LD states a publication date of 2026-03-16 (no visible date in body text).
Why it matters, evidence and limitations
- Why it matters
- Fills the thin U.S. municipal stormwater measured-results gap with an AU local-government CCTV+AI capital-prioritisation example. Transferability limited by Australian standards and vendor authorship; useful as a pattern for stormwater CIP screening with human capital oversight.
- Evidence and measured results
- Opened Vapar case-study page; metrics and budget figures are vendor/customer claims on that page without third-party evaluation in-source.
- Limitations and uncertainty
- Vendor case study dated 2026-03-16 in Article JSON-LD; older contextual evidence, not a new November deployment; AU context; no accuracy trial; $7.7M budget figure not independently verified here.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.
- Public-sector association guidance
- Practitioner guidance or an association-supplied case; not independent outcome evidence.
- Independent research
- Research conducted outside the implementing organization.
- Academic research
- Research produced through an academic institution or peer-reviewed venue.
- Standards or public-body guidance
- Normative or advisory guidance from a standards body or public institution.
- Vendor claim
- A supplier-provided assertion that has not been upgraded to independent evidence.