Claude is a Game Changer for Global Sustainability
Why corporate sustainability data is broken, why the regulatory window is closing, and what AI can do about it.
Most large companies now publish a sustainability report. The problem is that many of those reports are incomplete in ways that matter.
Scope 3 emissions – the upstream and downstream footprint from a company’s value chain — typically represent the majority of a company’s total carbon impact. McKinsey estimates the figure at around 90 percent for most industries, and peer-reviewed research confirms that for many corporations, Scope 3 accounts for 75 percent or more of their total emissions profile. Yet a 2021 study published in Nature Communications, applying a standardised methodology to 56 technology companies, found that corporate disclosures omit roughly half of total emissions — with individual company footprints understated by a median factor of four. Research from BCG puts the problem plainly: over 90 percent of companies cannot accurately measure their emissions.
The reports exist. The measurement often does not.
I have spent 25 years in the operational middle of this problem: co-creating Cisco EnergyWise, building Amazon’s global sustainable buildings practice, and conducting energy and carbon audits across more than 70 data centres. What I found, consistently, is that organisations were reporting on sustainability before they had measured it. This series is about fixing that and about how AI, specifically Claude, changes what is now possible.
Scope 1 and 2 are the visible tip. For most companies, Scope 3 is everything below the waterline.
What a baseline actually requires
The GHG Protocol Corporate Standard is the globally dominant framework for corporate emissions accounting, developed jointly by the World Resources Institute and the World Business Council for Sustainable Development defines the requirements for a defensible carbon inventory: a defined organisational boundary, a chosen consolidation approach (equity share or operational control), a baseline year, and documented methodology consistent enough to allow year-on-year comparison.
That last requirement is where most programmes break down. A carbon inventory is only meaningful if the methodology is stable and the boundary is locked. In practice, companies grow, acquire, and divest. Facilities come in and out of scope. When a 2030 target was set in 2022, did it include facilities acquired in 2024? If not, reported reductions may simply reflect what the company chose to count, not what actually changed.
I have seen organisations report 15 percent emissions reductions on paper while their actual footprint grew. Not because anyone was being dishonest, but because the boundary had never been formally documented and maintained.
The three most common measurement failures
Scope 1 (direct combustion) and Scope 2 (purchased electricity) are relatively tractable. Scope 3 – covering 15 upstream and downstream categories including purchased goods and services, employee commuting, business travel, and the use of sold products — is where accuracy collapses. Under current global frameworks, Scope 3 reporting remains largely voluntary for most companies outside the EU, and the dominant calculation method (spend-based estimation) is widely acknowledged by the GHG Protocol itself to be inaccurate. The Protocol is currently revising the Scope 3 Standard specifically to address data quality concerns, with a progress update published in March 2026. Reporting Scope 1 and 2 as a complete picture is like measuring an iceberg by what is above the waterline.
A baseline is only comparable over time if the organisational boundary is explicitly documented and consistently applied. The GHG Protocol requires companies to recalculate base year emissions when significant structural changes occur (acquisitions, divestments, outsourcing), but in practice this recalculation is frequently omitted or applied inconsistently. The result is that year-on-year comparisons that appear to show improvement may instead reflect boundary changes rather than actual decarbonisation.
Most large-organisation carbon inventories are produced through a manual annual process: a small team chasing utility bills, procurement records, and supplier disclosures across dozens of countries, reconciling inconsistent formats, filling gaps with estimates, and publishing a number. When that team changes, institutional knowledge leaves with them. When the methodology shifts, there is no audit trail. When a regulator asks a question, the answer takes months. This is not a competence problem. It is an infrastructure problem. The data systems sustainability teams depend on were built for financial reporting, not carbon accounting. The gap is bridged, year after year, by individual effort.
Why the regulatory window is closing
For most of the past two decades, sustainability reporting was voluntary for companies outside the EU. That is changing, though the pace varies by jurisdiction.
The EU’s Corporate Sustainability Reporting Directive (CSRD) requires large EU companies and non-EU companies with significant EU revenue to report Scope 1, 2, and 3 emissions under the European Sustainability Reporting Standards (ESRS). The first wave of companies reported for financial year 2024. A December 2025 Omnibus simplification package has narrowed the scope, concentrating requirements on companies with more than 1,000 employees, but the core emissions disclosure obligations remain.
In the United States, the federal picture is less settled. The SEC’s climate disclosure rule, finalised in March 2024, is currently stayed pending litigation, and the Commission ended its defence of the rule in March 2025. However, California’s SB 253 requires companies with over $1 billion in annual revenue doing business in California to disclose Scope 1 and 2 emissions from 2026 and Scope 3 from 2027. Given California’s economic scale, this reaches a substantial portion of major US companies regardless of federal action. Australia’s mandatory climate reporting framework began phasing in for large listed entities in 2025.
The direction of travel is clear even where the timeline is contested: auditable, third-party-verified carbon data is becoming a legal requirement in major markets, not a differentiator.
The infrastructure gap: most sustainability teams are managing a data engineering problem with spreadsheets and institutional memory.
What Claude changes
Claude changes the leverage equation at most of those steps. What I have found, working with Claude on sustainability data problems, is that it is genuinely useful not because it knows your data. It does not, but because it can reason across complex frameworks, translate between reporting schemas, identify gaps in a methodology, and help a non-specialist understand what they are looking at. The GHG Protocol Scope 3 Standard runs to 180 pages. CSRD’s ESRS E1 is not light reading. Claude can navigate both in the same conversation.
Concretely, Claude can assist with:
- Mapping existing data fields to GHG Protocol Scope 1, 2, and 3 categories and identifying what is missing
- Cross-referencing a company’s current methodology against CSRD ESRS E1 or California SB 253 requirements and flagging compliance gaps
- Extracting relevant emissions data from utility records, procurement reports, or supplier disclosures
- Drafting methodology statements that document calculation approaches, emission factors used, and data quality limitations
- Identifying where spend-based estimates are in use and suggesting primary data sources that would replace them with actuals
None of this replaces a sustainability professional who understands the substance. What it does is give that professional, or a small team with constrained capacity, a meaningful multiplier on throughput.
At SafetyCulture, we ran a structured evaluation of AI engineering tools across a team of 221 engineers: Copilot, Cursor, Windsurf, and Claude. Claude was the tool of choice, and the reason was consistent — it was the best at reasoning across complex, multi-domain problems. Sustainability data is exactly that kind of problem.
Where to start
If you are thinking about your own baseline, the practical starting point is more straightforward than it might appear.
- Establish and document your organisational boundary. Decide whether you are using equity share or operational control consolidation under the GHG Protocol, write it down, and date it. Every subsequent year’s calculation depends on the consistency of that decision.
- Start with what you can measure. A baseline built on 70 percent actual data and 30 percent documented, defensible estimates is more useful than no baseline, as long as the methodology distinguishes between the two and commits to improving the ratio over time.
- Build for repeatability, not heroism. If your data collection process depends on one person knowing which contacts to reach in which countries, it will fail when that person leaves. The process needs to be documented and reproducible by someone who was not there when it was built.
The infrastructure question, how to make this repeatable at scale, how to handle data that does not communicate over standard protocols, how to close the Scope 3 gap is what the rest of this series addresses.
Rob Aldrich
Technology and sustainability executive based in Sydney, Australia. Co-creator of Cisco EnergyWise. Led Amazon’s Global Sustainable Buildings practice. Author of IP-Enabled Energy Management (Wiley/IEEE Press). Former iMasons Sustainability Chair and ISO standards contributor.
robaldrich.com · LinkedIn · @googlenut
Sources
- GHG Protocol Corporate Accounting and Reporting Standard, World Resources Institute / WBCSD
- GHG Protocol Scope 3 Standard — Phase 1 Progress Update, March 2026
- Karstensen et al. (2021), “Harmonizing corporate carbon footprints,” Nature Communications
- Hettler & Graf-Vlachy (2024), “Corporate Scope 3 Carbon Emission Reporting,” Business Strategy and the Environment
- McKinsey & Company, “What are Scope 1, 2, and 3 emissions?” (2024)
- EU Corporate Sustainability Reporting Directive (CSRD), European Commission
- California SB 253, Climate Corporate Data Accountability Act
- Clarity AI, “Carbon Reporting Trends: Has Global Progress Stalled?” (2024)
