Skip to content
GLOBALANCE INTELLIGENCEThe economy ahead.
This Week Index 28/100 Sign in
Money & Investments › GI 100 Index
The Great Value ShiftInvestment ThemesGI 100 IndexGI PortfolioStart Future Check
INVESTMENT STRATEGY · THE AI FOOTPRINT

How does a company approach this technology?

The classic Globalance Footprint® measures the ecological and social footprint. The GI 100 required something different: a footprint of technology itself. Each of the roughly one hundred Business receives the same question. The answer rests on documented observations, not on judgement.

Measurement covers five dimensions. Governance: is there oversight and named accountability? Transparency: does the Business disclose where it deploys AI? People and labour: retraining, fair transitions, and fairness towards those affected. Safety: bias, data protection, and safeguards against misuse. Actual behaviour: documented incidents and proceedings over the past 36 months. Each dimension receives one of four words. The overall verdict is then calculated from those ratings using fixed weights.

The four ratings mean the same thing at every level. Strong means: published, specific, verifiable evidence, and no documented controversy. Solid means: substantiated structures, but lacking consistent detail. Limited means: principles without concrete evidence, or the harm rule applies. Unclear means: no disclosure found. That is a transparency gap, not misconduct, and is never counted as a negative.

What this yields

The framework covers all 96 securities in the universe. The same standard, the same run, with sources cited. As of July 2026.

5
49
37
5
5strong

Published, concrete evidence and no documented controversy.

49solid

Documented structures, but without consistent specificity.

37limited

Principles without concrete evidence, or the harm rule applies.

5unclear

No disclosure found. A transparency gap, not misconduct.

No company scores consistently «strong»Disclosure on AI conduct is still young across the market. That is a finding, not a flaw in the method. It makes a top rating something earned rather than something expected.
Conduct outweighs stated intentFor several prominent companies with strong governance the harm rule applies: their rating cannot be better than “limited”. A good set of principles does not outweigh a documented controversy.
Silence is not a clean bill of healthA company that discloses nothing is rated «unclear». Handled neutrally, never counted as a negative. But «no controversy found» alone is not enough to reach the top.

What this is for, and what it is not. Today the Globalance AI Footprint is a disclosure tool. It explains how the scrutiny is conducted and makes differences visible. It affects not security selection, not weighting. That is a deliberate first step. With experience and data, it may become an exclusion criterion in time.

The same question, two perspectives

The ethical question differs across the three roles in the index. Companies that build or sell AI as a product face different questions from those that use it. The same framework applies, but with different weights.

Enablers and innovatorshow is AI built?Model safety, transparency and explainability, data handling, protection against misuse and dual-use. Here, the safety of the systems being built carries the greatest weight.
Adoptershow is AI deployed?Treatment of the workforce, fairness in decisions affecting individuals, transparency with customers. Here, how people are treated carries the greatest weight.

Five dimensions

1
Governance & Accountability
Is there oversight, named accountability, and published principles?
both equal
2
Transparency & Explainability
Does the company disclose where and how it uses AI?
both equal
3
People & Labour
Retraining, fair transitions, and fairness towards those affected.
stronger among adopters
4
Safety & Loss Prevention
Bias control, data protection, and safeguards against misuse.
weighted more heavily for enablers and innovators
5
Controversies & real-world conduct
Documented incidents, proceedings, and regulatory breaches over the past 36 months.
double weight

The brand on the right refers to this as Weight of the dimension, not of a company's outcome. «Equal for all»: the dimension carries the same weight across all three roles. «Stronger for adopters» and «stronger for enablers and innovators»: it carries more weight for those roles because the question is sharper there. A company that deploys AI is responsible for its workforce's transition. One that builds or sells AI is responsible for the safety of the systems it creates. «Counts double»: documented behaviour counts twice as much as any other dimension. A verified controversy eclipses any policy paper.

Four levels: one word per dimension

strong
Published, concrete, verifiable evidence or documented behaviour.
solid
Substantiated structure, but partly lacking specificity or relevance to the product.
limited
Principles or statements of intent only, without concrete evidence.
unclear
No disclosure found in the course of research.

The most important principle: «unclear» is not negative. Missing disclosure is a transparency gap, not misconduct. It is excluded from the calculation rather than counted as a negative. That is fair to companies that simply publish less in English.

Two safeguards against misjudgement

The harm ruleIf a documented incident is classified as serious, the overall verdict is capped at «limited», regardless of how sound the stated principles appear.
Disclosure is a prerequisite, not a bonusAny entity that discloses nothing on more than a third of the grid is likewise rated no higher than 'limited'. Above half, the overall verdict is 'unclear'.The overall verdict is calculated by the system, not the language model. The five ratings are combined into a single score using fixed weights, with fixed thresholds. Every overall verdict is therefore reproducible and verifiable. The same inputs produce the same result, every time.Selected profiles in detailA profile appears here only once it has been approved for publication. Without approval, no verdict appears, not even a positive one.
SIX PROFILES · APPROVED
Siemenssolid
● Innovators · Germany · Automation & Robotics
WEIGHT
3,3 %
max. 5%
CONVICTION
63
As at 8 July 2026

Solid, though less product-specific disclosure; co-founder of the Charter of Trust and published Responsible AI principles. No documented AI controversies within the scope of this research.

Governance & Accountability
solid
Transparency & Explainability
limited
People & Labour
unclear
Safety & Loss Prevention
solid
Controversies & real-world conduct
strong
Netflixsolid
● Adopters · USA · Communications & Media
WEIGHT
2,8 %
max. 5%
CONVICTION
—
not yet collected

Broad AI user (recommendations for >300m users, GenAI in productions). First published GenAI production principles in 2025, including a consent clause; shareholder/union tensions and a GDPR fine.

Governance & Accountability
solid
Transparency & Explainability
solid
People & Labour
solid
Safety & Loss Prevention
solid
Controversies & real-world conduct
limited
Toyotasolid
● Adopters · Japan · Industry & Manufacturing
WEIGHT
1,7 %
max. 5%
CONVICTION
—
not yet collected

Published «AI Principle» (human-centred, fair, transparent). Active AI enablement of factory employees (10,000 employee-created models, no facial-recognition deployment); one pending US telematics data action.

Governance & Accountability
solid
Transparency & Explainability
solid
People & Labour
strong
Safety & Loss Prevention
solid
Controversies & real-world conduct
limited
Marvellstrong
● Enablers · USA · Semiconductors & computing power
WEIGHT
1,5 %
max. 5%
CONVICTION
—
not yet collected

Custom AI accelerators and networking chips. In 2024/25, established its own responsible-AI governance framework, with published principles, a dedicated committee, and board-level oversight.

Governance & Accountability
strong
Transparency & Explainability
solid
People & Labour
solid
Safety & Loss Prevention
strong
Controversies & real-world conduct
strong
Alphabetsolid
● Innovators · USA · Software, Agents & Platforms
WEIGHT
0,8 %
max. 5%
CONVICTION
76
As at 8 July 2026

Comprehensive, published AI governance and annual progress reports. At the same time, a documented governance retreat that has drawn external criticism: the removal of self-imposed restrictions on weapons and surveillance applications (02/2025).

Governance & Accountability
strong
Transparency & Explainability
solid
People & Labour
unclear
Safety & Loss Prevention
solid
Controversies & real-world conduct
limited
Infosyssolid
● Adopters · India · Technology Adopters
WEIGHT
0,7 %
max. 5%
CONVICTION
—
not yet collected

AI as a provider (Topaz) and internally. Most mature governance in the pilot: Responsible AI Office (2024), ISO/IEC 42001 certification, NIST AISIC member; one labour-related trainee controversy (2025, no documented link to AI).

Governance & Accountability
strong
Transparency & Explainability
solid
People & Labour
solid
Safety & Loss Prevention
solid
Controversies & real-world conduct
limited

6 of 97 holdings approved. The full profile for each holding is available in the company list.

What we are not claiming

No ideological screeningWhat is measured is disclosure and documented behaviour, not a company's stated position.
No sustainability labelThe Globalance AI Footprint does not replace the Globalance Footprint® and is not an overall ESG rating.
Observation, not judgementThe same framework, the same sources, the same reference date for every security. Evidenced rather than assessed.Evidenced observations drawn from public sources · As at July 2026 · No conclusive ethical assessment, no sustainability label, and no investment recommendation.
→ Back to the GI 100 portfolioThe securities to which the framework is appliedImportant Notice. This content is provided for informational purposes only and does not constitute an offer or solicitation to buy or sell any financial instrument, nor does it constitute investment advice or a recommendation. GI 100 is a rules-based strategy index of Globalance Bank; the associated investment product is in preparation. There is no real track record. Historical figures are based on a point-in-time model calculation (in USD, before transaction and implementation costs) and are not indicative of future results. Investments involve risks, including the loss of capital invested. The final product documents and applicable legal and regulatory requirements (including FinSA) shall solely prevail. © Globalance Bank AG.

What comes next