AI Footprint

Archives
Log in
Subscribe
September 25, 2026

AI Footprint: Alberta’s 1,200 MW limit, Southeast AI skills, and CSF GenAI drafts

Photoreal editorial desk still life under soft daylight: open notebook with handwritten notes, printed research pages, laptop edge, coffee cup, reading glasses, and a fountain pen — warm workspace atmosphere for the September 25 AI Footprint edition

Friday, September 25, 2026 · Daily edition

Alberta’s filled 1,200 MW data-centre limit, Southeast AI-skill geography, and a CSF GenAI comment clock

Today’s ledger moves from a completed Alberta large-load allocation to where AI skills show up in Southeast job ads, a new NIST draft on using AI to write CSF 2.0 profiles, a WHO/Europe dialogue on responsible health AI, and a UNESCO teacher-literacy webinar that treats media literacy as the core of AI training.

Alberta’s interim 1,200 MW large-load limit is fully allocated — two data-centre contracts take the whole pie

What happened. Alberta Electric System Operator says Phase 1 of its Large Load Integration program is complete. The interim connection limit of 1,200 MW is fully spoken for under two executed load contracts: GLDC Load at 970 MW and Keephills Data Centre Phase I at 230 MW. Those projects continue through the normal connection process. Every other large-load request — including extra megawatts for the same sites — moves to Phase 2, which is building a longer-term framework across connections, planning, operations, markets, tariff, and reliability. Phase 2A is labeled a bring-your-own-generation path. A separate guide for transmission-connected data centres was posted 12 June 2026. This is a provincial ISO program-status page, not a U.S. RTO stakeholder deck and not a national electricity forecast.

What to watch. After a week of U.S. grid clocks and deposit ladders, the environment beat is a finished allocation: Alberta’s interim megawatts are spoken for, and everything else waits on a BYOG Phase 2 path — program design, not another national power curve.

Read the AESO Large Load Projects page →

Atlanta Fed: Georgia AI-skill postings hit 2.93% — and Atlanta takes more than four-fifths of the state’s AI ads

What happened. Federal Reserve Bank of Atlanta Workforce Currents 2026-04 maps Lightcast job postings that include at least one AI or machine-learning skill across six Southeast states from 2010 through 2025. By December 2025, Georgia’s share of postings that require AI skills was 2.93%, just under the national average near 3%, and Georgia leads the six-state region. AI-related ads are still only about 2–4% of total demand in each state, but the share has risen steadily and accelerated in recent years. Inside Georgia, the Atlanta metro accounts for more than four-fifths of statewide AI postings. Most of those ads ask for only one to four AI skills rather than a long specialist stack, which the authors read as AI folding into ordinary roles. Larger metros show higher AI specialization with lower employer concentration; smaller metros concentrate demand among fewer employers. This is a geography of skill demand — not bank in-house AI hiring intensity and not a layoff census.

What to watch. After bank hiring proxies and national vacancy studies, the jobs beat is where AI skills actually appear in Southeast postings — a modest share of ads, heavily metro-concentrated, mostly as a few skills inside broader jobs.

Read the Atlanta Fed Workforce Currents report →

NIST SP 1353 draft: using AI to draft CSF 2.0 profiles — comments open through 15 October

What happened. NIST published the initial public draft of Special Publication 1353, a quick-start guide for using artificial intelligence to help with Cybersecurity Framework 2.0 analysis and reporting. Comments run through 15 October 2026 at 11:59 p.m. NIST wants feedback on the guide and the supplied prompts, not on the fictional organizational documents used as examples. The draft sketches three notional use cases: AI-assisted review of policy and risk governance against CSF 2.0 outcomes; drafting an organization current-state profile from artifacts and interview notes; and drafting a target-state profile from internal and industry references. The cases are illustrative, not a required assessment method. This is a draft on paperwork assistance — separate from the AI data-center security draft whose comment window closes today, and not a Binding Operational Directive.

What to watch. On the same day the AI campus security draft closes, NIST opens a different lane: how practitioners may use AI to write CSF profiles — assistance for framework paperwork, not a new threat catalog.

Read the NIST SP 1353 draft page →

WHO/Europe: a 22-page Knowledge Community report on responsible AI in health

What happened. WHO’s Regional Office for Europe published a 22-page report from a five-week structured online dialogue held October through December 2025 with clinicians, researchers, policy-makers, and digital-health experts from across the Region and beyond. The report names barriers that keep coming up: fragmented and biased datasets, governance gaps, unclear accountability, and weak AI literacy. Enablers named in the same pages include strong governance, codesign with patients and clinicians, transparency, and equitable access. The document is meant to inform policy and guide health-system adoption. Limits are plain: regional dialogue proceedings, not a trial result and not a bedside authorization.

What to watch. After note-wording safety experiments and multi-agent discovery headlines, the health beat is how a regional knowledge community frames governance and literacy as the binding constraints on clinical AI — deliberation, not a deployment green light.

Read the WHO/Europe Knowledge Community report →

UNESCO IITE: media literacy as the cognitive core of teacher AI training

What happened. UNESCO’s Institute for Information Technologies in Education recapped the third 2026 IFAP Information Literacy Working Group webinar, held 24 September 2026, on integrating media and information literacy with AI capacity-building for lifelong learning. Speakers were named from Kazakhstan, Pakistan, the Philippines, and UNESCO IITE. The recap maps AI competencies onto MIL competencies — prompt work overlapping how people name information needs; AI-supported research overlapping question framing, source selection, synthesis, and verification. Kazakhstan’s path is named explicitly: a National AI Concept for 2024–2029 plus curriculum objectives from primary through upper secondary. The line that travels: “Access does not equal literacy.” This is a teacher professional-development webinar recap, not a higher-ed uptake census and not a learning-outcomes print.

What to watch. After regional ethics forums and higher-ed strategy surveys, the education beat is how teacher training treats media and information literacy as the thinking core of AI skills — capacity, not another enrollment or test-score dump.

Read the UNESCO IITE webinar recap →

Also in today’s ledger

• A finished 1,200 MW provincial allocation is not a national terawatt-hour path — and not yesterday’s New York 50 MW stakeholder design.

• Southeast AI-skill geography is not a bank AI-hiring proxy — and a CSF GenAI quick-start is not an AI data-center security draft.

• Regional health-AI dialogue is not a triage experiment — and teacher MIL↔AI mapping is not an 87%/26% higher-ed census.

Full ledger

This is the short version.

The complete source-linked ledger is on AI Footprint.

Open today’s full AI Footprint edition →

Open the dated September 25 edition →

Don't miss what's next. Subscribe to AI Footprint:
← Newer AI Footprint: data-center water gallons, China hybrid AI jobs, and EU Board cybersecurity Older → AI Footprint: NYISO large loads, bank AI hiring, and NIST data-center security
Powered by Buttondown, the easiest way to start and grow your newsletter.