How do the AI winners deploy their technology?
Those who profit from AI bear responsibility for what it does. The Globalance AI Footprint assesses how seriously the companies in the GI 100 Index take that responsibility.
Download the summary (PDF, 7 pages, German) ›Why Globalance rates this
Since its founding, Globalance has assessed not only the return but also how companies affect the economy, society and the environment. For years we have used the Globalance Footprint to do so.
Artificial intelligence raises a new question: how do companies handle a technology that replaces people, makes decisions about them and shapes their behaviour? To address this, we developed the Globalance AI Footprint. We publish it here, company by company, with documented observations.
How we assess.
Three pillars, one yardstick for all, public evidence only.
Responsibility
Who has oversight? Is responsibility named? How does the company treat people and work?
Safety
Does it protect against harm, misuse and bias?
Transparency
Does it disclose where and how AI is used?
The width of each pillar shows its weight. How three pillars become a rating ›
How the ratings are distributed
Every company receives a rating on one of four levels, each justified with evidence. In 2026 no company reaches «strong» yet.
Result of the assessment by the Globalance research team under the published methodology. Not a certification or quality label. How we assess ›
What the results show.
We applied the rulebook to all 100 companies in the GI 100 Index: each company assessed individually, with the same yardstick and with sources. As of 3 November 2026.
Published, concrete evidence and no documented controversy.
Documented structures, but without consistent specificity.
Principles without concrete evidence, or the harm rule applies.
No disclosure found. A transparency gap, not misconduct.
Three findings
No company has yet reached «strong»
No company in the GI 100 reaches «strong» today. 35 reach «solid», 65 «limited». Disclosure on how AI is handled is young across the market. That is a finding, not a flaw in the method, and it makes the top an earned goal.
Conduct outweighs stated intent
For 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.
Security over accountability
At 86 of the 100 companies, security rates at least «solid»; accountability does so at only 28, and transparency at 21. Companies protect their systems more effectively than they govern their AI.
How does a particular company handle AI?
The profile is published for all of the 100 companies in the GI 100. Assessed under the rulebook: responsibility 40%, safety 40%, transparency 20%, with around 40 searches and at least 15 sources read per company. One click opens the profile with all evidence.
Click a row to open the short assessment. The full profile with all evidence is one level deeper.
| Company ▲ | Country | Responsibility | Safety | Transparency | Rating | |
|---|---|---|---|---|---|---|
| ABB Ltd | Enablers | Switzerland | limited | solid | limited | limited |
ABB uses artificial intelligence widely in industrial products and in its own operations, and in its code of conduct it commits to safe, lawful and ethical use under human control. What has been confirmed by regulators is mainly its rules for protecting personal data, while no AI policy of its own, no clear oversight by top management and no information on the limits of its AI products were found. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Switzerland | ||||||
| Accenture PLC | Adopters | Ireland | limited | solid | limited | limited |
Accenture has assigned oversight of AI risks to a board committee and can show independent certificates for protecting client data, but describes its rules for its own AI only in general terms and without verifiable figures. An ongoing US Justice Department investigation into security statements by a subsidiary and a data leak in summer 2026 cloud the picture on security. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · Ireland | ||||||
| Advantest Corp | Enablers | Japan | limited | limited | limited | limited |
Advantest, a Japanese manufacturer of test equipment for computer chips, uses artificial intelligence in its own analytics offerings and in operations, but discloses hardly any rules, responsibilities or testing procedures for it. In a cyberattack in February 2026 personal data was stolen, and those affected were notified only in October 2026. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Japan | ||||||
| Adyen NV | Adopters | Netherlands | limited | solid | solid | solid |
Adyen uses artificial intelligence mainly for fraud detection and to improve payments, and in its product documentation shows merchants which signals led to a risk assessment and where detection reaches its limits. The company cites its own principles and a working group on the ethics of AI and data, but has not published these principles, and affected customers are told little about how they can contest automatic rejections. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryAdopters · Netherlands | ||||||
| Airbnb Inc | Adopters | USA | limited | solid | solid | solid |
Airbnb uses artificial intelligence widely, for example in customer service and in screening risky bookings, and informs users about these uses and about the option to object to the use of their data for training. Clear published rules and responsibilities for handling artificial intelligence are lacking so far, and no independent testing of the systems was found. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryAdopters · USA | ||||||
| Alibaba | China | limited | limited | limited | limited | |
Alibaba develops its own freely available AI models with Qwen and Wan and cites principles and safety tools, but independent assessments criticise missing safety rules, weak defences against targeted attacks and sparse information in the model descriptions. In addition, its subsidiary AliExpress was fined in South Korea for passing customer data to foreign sellers without adequate protection, and the open video models are demonstrably misused for non-consensual content. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryInnovator · China | ||||||
| Alphabet | USA | solid | solid | solid | solid | |
Alphabet discloses detailed safety reports, principles and testing procedures for its AI models and has important AI services certified to an international standard for AI management systems. At the same time, the company faces several lawsuits and investigations over false or harmful chatbot answers and has dropped its earlier pledge not to develop AI for weapons and surveillance. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryInnovator · USA | ||||||
| Amazon | USA | solid | solid | solid | solid | |
Amazon has published traceable testing, security and labelling procedures for its own AI models and cloud services and has part of them externally certified. At the same time, lawsuits are pending over facial recognition in Ring cameras and in virtual try-on, and in France a fine for storing data on warehouse workers for too long was upheld in court. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryInnovator · USA | ||||||
| AMD | Enablers | USA | limited | solid | solid | solid |
AMD supplies chips and open software for artificial intelligence, publishes its own models including training data and fixes reported security flaws through an established disclosure process, but serious flaws in chips and software have occurred repeatedly over the past three years. Principles and a body for responsible AI have been announced, but AMD discloses little about how they are implemented and monitored in practice. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryEnablers · USA | ||||||
| Amgen Inc | Adopters | USA | limited | solid | limited | limited |
Amgen has assigned oversight of artificial intelligence to a board committee and an internal expert council and describes an extensive programme to protect its IT systems, but publishes no binding rules on how AI applications are tested and approved. The company discloses little about the limits of individual AI applications, and in July 2026 patient data was stolen in an attack on cloud storage, although a link to AI is not established. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Amphenol Corp | Enablers | USA | limited | solid | limited | limited |
Amphenol makes connectors and cables that are in strong demand in data centres for artificial intelligence, but discloses little about how the company itself uses AI and which rules apply. To protect its IT systems, it describes oversight by the board and regular testing, and no cyber incidents became known during the review period. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Anthropic | USA | solid | limited | solid | solid | |
An unusually well-anchored governance structure (public benefit corporation, Long-Term Benefit Trust, ISO 42001) and very detailed system cards support a “solid” rating. At the same time, the founders hold 50.1% of the votes on shareholder matters through a separate entity; for a company heading to an IPO this is unusual and contested, as future shareholders will have little say. The USD 1.5bn copyright settlement over pirated books, cyber incidents involving its own models and further pending lawsuits keep the safety pillar at “limited”. Responsibility · 40 %solid Safety · 40 %limited Transparency · 20 %solid Role · CountryInnovator · USA | ||||||
| Aon plc | Adopters | United Kingdom | solid | solid | limited | solid |
Aon has assigned oversight of risks from the use of artificial intelligence to a board committee and contractually commits to its clients not to use their data to train such systems. What remains open is how fair the group's computer-based hiring tests are: a civil rights organisation has challenged them with the US authorities as discriminatory, and Aon publishes no test results on them. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · United Kingdom | ||||||
| Applied Materials | Enablers | USA | limited | solid | limited | limited |
Applied Materials demonstrably protects its IT under a recognised, externally certified standard. The company discloses only in general terms how it governs, tests and explains artificial intelligence in its products and internally; clear responsibility for AI at board or management level is not documented. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| AppLovin Corp | USA | limited | limited | solid | limited | |
AppLovin discloses who on the board is responsible for dealing with artificial intelligence and data protection and which data its advertising AI uses, but publishes no policy of its own and no testing of these systems. Independent analyses and several pending lawsuits in the US and the Netherlands accuse the company of collecting data, including from children, without adequate consent, which AppLovin denies. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %solid Index assessment: limited Role · CountryInnovator · USA | ||||||
| Arista Networks Inc | Enablers | USA | limited | solid | limited | limited |
Arista builds network technology for AI data centres and cites an internal policy and a management committee for dealing with AI, but their content is not public. Arista has the security of its cloud management audited externally every year, but in 2026 it had to report several security flaws in its network software that were exploited by attackers. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| ARM Holdings PLC | Enablers | United Kingdom | limited | solid | solid | solid |
Arm designs the computing building blocks on which a large share of artificial intelligence in phones, cars and data centres runs, and has an established process for reporting and fixing security flaws with oversight by the board, even though flaws in graphics drivers have already been exploited by attackers. For the responsible use of artificial intelligence, Arm refers to an internal policy and a dedicated office, but does not disclose their content or effect. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryEnablers · United Kingdom | ||||||
| Asia Vital Components Co Ltd | Enablers | Taiwan | limited | limited | limited | limited |
Asia Vital Components makes cooling systems for AI data centres and has recently started using its own AI tools, but discloses neither rules nor responsibilities or controls for them; on the protection of its data and systems, there are mainly the company's own statements and plans. No court proceedings, fines or confirmed security incidents were found in the period examined. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Taiwan | ||||||
| ASML Holding | Enablers | Netherlands | limited | limited | limited | limited |
ASML has published principles for the responsible use of artificial intelligence and has IT security monitored regularly by its supervisory body, but discloses little about how it actually tests and manages AI in its systems and internally. A former employee was convicted in 2025 for passing confidential documents to Russia, and deliveries to military-linked institutions in China have been criticised, without authorities having found a violation. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Netherlands | ||||||
| AstraZeneca PLC | Adopters | United Kingdom | limited | solid | limited | limited |
AstraZeneca has published principles for the ethical use of data and AI and trains tens of thousands of employees in working with AI, but provides little evidence of how the rules are checked and monitored in day-to-day practice. On the protection of data and systems, the company reports no material security incidents for 2024 and 2025, while in China criminal proceedings are under way against a subsidiary over the unlawful collection of patient data. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · United Kingdom | ||||||
| Automatic Data Processing | Adopters | USA | solid | solid | solid | solid |
ADP has anchored oversight of artificial intelligence with its board, describes formal reviews of new AI applications before launch in its corporate responsibility report, and explains unusually openly how its tool for assessing job applications works and that an independent audit found no disadvantage to any group. For the many new AI assistants in payroll and HR processes, however, the company has so far provided hardly any verifiable information on testing, limitations and the data used. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryAdopters · USA | ||||||
| Baidu Inc | China | limited | limited | limited | limited | |
Baidu discloses structures for cybersecurity and data protection and publishes open models with technical reports, but verifiable information on safety testing, the limitations of its AI and the handling of complaints is missing. A Chinese court ruled with final effect that an AI answer in Baidu's search falsely portrayed a lawyer as a convicted criminal, and in March 2026 numerous robotaxis of the company came to a standstill on the roads of Wuhan because of a system failure. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryInnovator · China | ||||||
| Bristol-Myers Squibb Company | Adopters | USA | limited | solid | limited | limited |
Bristol Myers Squibb has anchored oversight of artificial intelligence with its board and two of its committees and describes a regularly externally assessed programme to protect its IT systems. However, the company has not published binding rules of its own for dealing with artificial intelligence, and it discloses very little about how individual AI applications, for example in drug safety or manufacturing, are tested and where their limits lie. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Broadcom Inc | Enablers | USA | limited | solid | limited | limited |
Broadcom supplies chips, networking technology and software for data centres running artificial intelligence and has a fixed procedure for reporting and fixing security flaws in its software, yet in recent years attackers have exploited several flaws in its virtualisation software. Broadcom discloses hardly any principles, responsibilities or reports of its own on the responsible use of artificial intelligence. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Cadence Design Systems | Enablers | USA | limited | solid | limited | limited |
Cadence describes the protection of its computer systems and data in specific terms and has it overseen regularly by the board, but discloses little about how its own AI tools for chip design are tested, governed and documented. No published policy or review procedure for the use of AI was found, and information on training data and limitations is missing for the new AI models. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| CBRE Group | Enablers | USA | solid | solid | limited | solid |
CBRE uses artificial intelligence widely, for example in an internal assistance platform for employees, in building operations and in property analysis, and has the risks of this technology overseen directly by the company's top supervisory body. The company publishes principles and rules of conduct for the use of AI, but no explanations of individual systems and no report of its own on how they are tested. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryEnablers · USA | ||||||
| Celestica Inc | Enablers | Canada | limited | solid | limited | limited |
Celestica builds servers, networking equipment and entire computer racks for large data centres that run artificial intelligence, and in 2026 expressly assigned oversight of AI risks to the board and its committees. The company discloses little about how it uses AI itself and under what rules; for the protection of its IT it describes regular reviews, and no cyber incidents became known during the period under review. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Canada | ||||||
| Ciena Corp | Enablers | USA | limited | solid | limited | limited |
Ciena supplies fibre-optic technology for AI data centres and offers software with AI assistants for network operations, but has published no rules of its own on the use of AI. The company describes a comprehensive cybersecurity programme; in summer 2026, however, several serious security flaws in its network software became known. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Comfort Systems USA | Enablers | USA | limited | solid | limited | limited |
Comfort Systems USA builds HVAC, electrical and modular plants, including for data centres, and is piloting a voice assistant using artificial intelligence that helps service technicians with troubleshooting and passes uncertain cases on to experts. For protecting its IT, the company describes oversight by the board and regular reviews, but publishes no rules of its own on the use of artificial intelligence. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Constellation Energy Corp | Enablers | USA | limited | solid | limited | limited |
Constellation mainly supplies electricity to AI data centres and itself uses purchased AI tools, for example to check sensors in nuclear power plants, where a fault was demonstrably detected and confirmed. Internally, the company requires every use of AI to be approved by a dedicated team, but discloses neither its AI rules nor the limits of the systems it uses. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| CoreWeave Inc | USA | limited | solid | limited | limited | |
CoreWeave rents out computing capacity for artificial intelligence to large technology companies and states that its cloud is audited against recognised security standards, but the audit reports are not public. The company has published hardly any rules of its own on the responsible use of AI or on protecting the platform against misuse by customers. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryInnovator · USA | ||||||
| Corning Inc | Enablers | USA | limited | solid | limited | limited |
Corning makes optical fibre and connectivity technology that is in high demand for data centres running artificial intelligence, and has its own use of AI overseen by a board committee. Its internal rules on AI are not public, and the company reports in more detail on protecting its IT systems than on its AI applications. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Deere & Co | Adopters | USA | solid | solid | limited | solid |
Deere uses artificial intelligence in agricultural machinery, for example for targeted weed spraying, for autonomous tractors and for a new digital assistant, and has explicitly assigned oversight of these risks to a board committee. However, there is no public policy for artificial intelligence and no safety evidence for the autonomous machines, and the company has not disclosed which model powers the assistant. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · USA | ||||||
| Delta Electronics Inc | Enablers | Taiwan | limited | solid | limited | limited |
Delta Electronics supplies power and cooling technology for AI data centres and uses AI in industrial automation and, through its subsidiary VIVOTEK, in video surveillance, but has not published any rules of its own on the use of AI. For the protection of its systems and products, the company presents audited certificates and clear reporting channels; in February 2026, however, systems of a foreign subsidiary were attacked. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Taiwan | ||||||
| Eaton Corporation | Enablers | Ireland | limited | solid | limited | limited |
Eaton uses artificial intelligence in products for power grids and buildings and in its own customer service, and its code of conduct calls for responsible use. Protection against cyberattacks and the handling of reported security vulnerabilities are well documented, whereas an AI policy of its own, clear oversight of AI by top management and information on the limits of its AI products are missing. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Ireland | ||||||
| Eli Lilly and Co | Adopters | USA | solid | solid | limited | solid |
Eli Lilly has a group-wide rule with prior approval for the use of artificial intelligence, oversight by the board and an audited information security programme, but discloses few figures, external audits or details on the limits of its own models. No harm caused by Lilly's AI is documented for the period; a law firm's inquiry into the sharing of usage data on the patient platform LillyDirect remains open. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · USA | ||||||
| Expedia Group | Adopters | USA | limited | solid | solid | solid |
Expedia uses artificial intelligence in customer service, search and pricing and against fraud, and describes an internal council that is meant to review risky applications before they are introduced. Users are told that answers may contain errors, but the third-party models used are not named, and no independent testing of the systems was found. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryAdopters · USA | ||||||
| FedEx Corp | Adopters | USA | solid | solid | limited | solid |
FedEx has assigned oversight of risks from artificial intelligence to a board committee, set up an internal policy and a review body for AI and trains its employees, but does not publish the policy itself. What remains open above all is how the company handles AI cameras at its sites and the sharing of this data with the police, on which it gives no information. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · USA | ||||||
| GE Vernova Inc | Enablers | USA | limited | solid | limited | limited |
GE Vernova builds artificial intelligence into software for power grids and power plants and commits in its code of conduct to safe and responsible use, but discloses neither its own AI policy nor information on testing and limitations of these functions. What is independently confirmed are mainly general security procedures for products, such as certified secure development in one business unit and orderly handling of reported security vulnerabilities. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Infineon Technologies | Enablers | Germany | solid | solid | limited | solid |
Infineon has a person responsible for artificial intelligence on its executive board, an internal rule with training and controls, and externally certified information security for its central systems, but discloses little about its own AI models. A vulnerability in its security chips discovered in 2024 has been partly fixed; for products that cannot be updated, only managing the risk remains. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryEnablers · Germany | ||||||
| Intel Corp | Enablers | USA | solid | solid | solid | solid |
Intel has described principles and bodies for responsible artificial intelligence up to board level and regularly discloses and fixes security vulnerabilities it finds, including in its own AI software. At the same time, there have been several security problems with processors, protection features and AI software over the last three years, as well as a pending class action over a processor flaw, and Intel discloses little about how its AI review procedures work in practice. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryEnablers · USA | ||||||
| Johnson & Johnson | Adopters | USA | limited | solid | limited | limited |
Johnson & Johnson has the use of artificial intelligence overseen by its board, trains employees widely and has audited procedures for information security, but publishes little on the review rules, limitations and test results of its AI applications. Lawsuits remain open over a navigation device for sinus procedures to which a former group subsidiary had added AI; no court has yet ruled on them. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Johnson Controls International | Enablers | Ireland | limited | solid | limited | limited |
Johnson Controls uses artificial intelligence in building control and security technology and has explicitly assigned oversight of it to a board committee; however, the company has not published documented binding internal rules, review or approval procedures for AI. The protection of its products against cyber attacks is documented by independent certificates and orderly security advisories; public information on testing of the AI functions, on their limitations and on the promised model descriptions is missing, and a lawsuit is pending over a data breach in 2023. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Ireland | ||||||
| JPMorgan Chase & Co | Adopters | USA | solid | solid | limited | solid |
JPMorgan Chase uses artificial intelligence widely, for example in fraud prevention and in voting at general meetings, and describes an independent review function that checks models before deployment, as well as board oversight of AI matters. Published AI principles and explanations for affected customers and applicants are largely missing, and a class action over an AI-assisted job interview is still pending before an appeals court. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · USA | ||||||
| Keysight Technologies | Enablers | USA | solid | solid | limited | solid |
Keysight makes measurement and test solutions that are in demand for building data centres for artificial intelligence, and at the end of 2025 introduced a group-wide policy with its own committee and mandatory training for the use of AI. For the protection of its IT systems, the company describes board oversight and certified security procedures, while it discloses little about the AI functions in its own products. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryEnablers · USA | ||||||
| KLA Corp | Enablers | USA | limited | solid | limited | limited |
KLA builds inspection and measurement equipment for chip manufacturing, uses artificial intelligence in some products and describes structured oversight of information security in its annual report. The company has not yet published its own rules for dealing with artificial intelligence, and in 2025 it reported unauthorised access to its systems. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Lam Research | Enablers | USA | limited | solid | limited | limited |
Lam Research demonstrably protects its IT according to a recognised, externally certified standard, but was publicly rebuked in 2025 by a committee of the US Congress because it did not deliver requested information on its sales to China on time. How the company governs, tests and explains artificial intelligence in its tools and internally is disclosed only in general terms. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Marsh & McLennan Companies | Adopters | USA | solid | solid | limited | solid |
Marsh & McLennan has assigned oversight of risks from artificial intelligence to a board committee and has new AI projects reviewed and approved by internal committees before deployment. Little is disclosed, by contrast, about how its own AI tools are tested and what data they are based on. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · USA | ||||||
| Marvell Technology | Enablers | USA | limited | solid | limited | limited |
Marvell develops chips for AI data centres and has a documented IT security programme and an officially certified security module. On the responsible handling of AI, the company has so far disclosed only general rules, for example for the internal use of AI tools. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Mastercard Inc | Adopters | USA | limited | solid | limited | limited |
Mastercard uses artificial intelligence mainly to detect fraud in the card payment network and to secure payments made by digital assistants, and has anchored oversight of this with its board. The company describes how new AI systems are reviewed and approved before deployment only in presentations and interviews, not in its formal reports; public descriptions of the individual systems and their limitations are missing, and no AI-related court cases or fines were found. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| MediaTek Inc | Enablers | Taiwan | limited | solid | solid | solid |
MediaTek protects its systems and chips in line with recognised security standards, discloses vulnerabilities regularly and publishes descriptions of its own language models as well as a safety model against fraudulent content. The company has so far only announced its own rules and responsibilities for dealing with artificial intelligence, and in the period examined no court cases, fines or incidents related to artificial intelligence were found. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryEnablers · Taiwan | ||||||
| Merck & Co | Adopters | USA | limited | solid | limited | limited |
Merck uses artificial intelligence widely in research, clinical trials, manufacturing and administration and had an important application for study reports tested by an independent assessor, but beyond general principles it discloses little about who in the company is responsible for AI and how new AI applications are reviewed and approved. No harm caused by Merck's own AI is documented in the period reviewed; two minor data protection cases concerned a website tracking lawsuit and a data incident at a service provider. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Meta Platforms | USA | limited | limited | solid | limited | |
Meta publishes detailed safety testing and model descriptions for its most capable AI models and has set up oversight by a board committee as well as named responsible officers. At the same time, independent investigations and court proceedings show serious failings in protecting children and vulnerable people in their dealings with Meta's chatbots, and remediation has so far been documented only in part. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %solid Index assessment: limited Role · CountryInnovator · USA | ||||||
| Microchip Technology | Enablers | USA | limited | solid | limited | limited |
Microchip makes chips and tools for artificial intelligence, and a board committee is now explicitly responsible for overseeing AI risks, but the company does not publish its own rules for dealing with AI. After a hacking attack in August 2024 in which employee data was stolen, the company describes expanded protective measures for its IT systems. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Micron Technology | Enablers | USA | solid | solid | limited | solid |
Micron has created its own governing bodies for the use of artificial intelligence, informs the board's oversight regularly about it and contractually prescribes how suppliers may deal with AI and Micron data; IT security follows a recognised US framework. The company discloses little about how its own AI applications in production are tested and where their limitations lie, and Micron faces, among other things, a class action over alleged price fixing for memory chips. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryEnablers · USA | ||||||
| Microsoft | USA | solid | solid | solid | solid | |
Microsoft has an extensive, publicly documented organisation for the responsible use of AI, with oversight by a board committee, annual transparency reports, extensive safety testing and external certifications. At the same time there are open proceedings and allegations, for instance on the use of the Microsoft cloud by the Israeli military, on repeated security flaws in the AI assistant Copilot and on pricing information challenged in Australia. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryInnovator · USA | ||||||
| Monolithic Power Systems | Enablers | USA | limited | solid | limited | limited |
Monolithic Power Systems supplies power supply chips, among other things for AI data centres, and documents the protection of its data and systems with external certificates, which however apply only to its Chengdu site. By contrast, it discloses little about how it steers and controls artificial intelligence in its own operations. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Morgan Stanley | Adopters | USA | limited | solid | limited | limited |
Morgan Stanley uses AI mainly as an aid for advisers and specialists and has clearly regulated oversight of technology and cybersecurity at board level. The bank has not published binding rules on how it tests and approves its AI tools before use, and discloses which data it uses for them only in general terms. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Nasdaq | Adopters | USA | solid | solid | solid | solid |
Nasdaq has anchored oversight of the use of artificial intelligence in the board's audit and risk committee, documents certified information protection and, under the supervision of the US securities regulator, discloses tests and monthly figures for its AI-powered order type. For the AI in its anti-money-laundering and market surveillance products, however, public test reports and explanations are missing, and several disruptions of its own trading systems are documented. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryAdopters · USA | ||||||
| Nokia Oyj | Enablers | Finland | limited | solid | limited | limited |
According to its own statements, Nokia has introduced a group-wide framework for artificial intelligence overseen by a board committee, but it does not publish the actual AI rules and review procedures. For its own AI products in telecom networks, too, the company discloses little about which models it uses and how it tests them, and in October 2026 a major network outage at Orange was attributed to a software fault in Nokia equipment. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Finland | ||||||
| Novartis AG | Adopters | Switzerland | limited | solid | limited | limited |
Novartis has published principles and a policy for the responsible use of artificial intelligence and describes a verifiable programme to protect its IT systems, but does not disclose who is responsible for AI at the top level or how individual AI applications are tested. Since March 2026 a class action has been pending in the US over the sharing of patient data via advertising tracking on product websites; no link to AI is alleged there. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · Switzerland | ||||||
| Nvidia Corp | Enablers | USA | solid | solid | solid | solid |
NVIDIA describes its rules for handling AI in detail, discloses information on training data and testing for many of its own models and has parts of its vehicle technology tested by independent bodies. At the same time, several lawsuits accuse the company of using books, videos and voice recordings for training without permission, and users of its open models as well as the US military face hardly any restrictions beyond the law. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryEnablers · USA | ||||||
| Oracle | USA | solid | solid | solid | solid | |
Oracle has had its AI services certified by an external body against a recognised standard for AI management, has the board oversee AI risks and explains in its product documentation how customer data is handled in AI services. Weighing against this are two large data breaches from older Oracle systems in 2025, including health data of almost 20 million people, and the lack of public reporting on testing and misuse protection of its AI. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryInnovator · USA | ||||||
| Palantir Technologies | USA | limited | solid | limited | limited | |
Palantir describes principles on privacy and human rights and holds a government security authorisation for its cloud, but discloses little about how it vets the use of its software by the military, police and immigration authorities. Several organisations and investors criticise these deployments, and reports link a Palantir system to a deadly strike on a school in Iran, which the company denies. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryInnovator · USA | ||||||
| Pfizer Inc | Adopters | USA | limited | solid | solid | solid |
Pfizer has principles for artificial intelligence, an internal group body for risk management and an information security programme audited by third parties, and its Health Answers health assistant openly shows sources, limits and unverified answers. Harm from Pfizer's own AI is not documented in the period; figures, external audits and information on its other AI applications are largely missing. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryAdopters · USA | ||||||
| Prysmian SpA | Enablers | Italy | solid | solid | limited | solid |
Prysmian has a group-wide policy for artificial intelligence approved by the board, with clear responsibilities and prohibited uses, as well as an externally certified information security system. Little is disclosed, however, about how reliably its own AI-supported monitoring product for power grids works and where its limits lie. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryEnablers · Italy | ||||||
| Regeneron Pharmaceuticals | Adopters | USA | limited | solid | limited | limited |
Regeneron has anchored oversight of artificial intelligence on its board in a dedicated committee and describes its protection of data and systems in specific terms in its mandatory reports to the securities regulator. How new AI applications are reviewed and approved and how the health and genomic data of millions of people are used for AI training, the company has so far disclosed only in general principles. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| RELX PLC | Adopters | United Kingdom | solid | solid | limited | solid |
RELX has published principles for the responsible use of artificial intelligence, set out the responsibilities for them in its annual report and described testing procedures for individual products. However, independent tests found errors in an AI research product, and lawsuits are pending against a subsidiary over its handling of personal data. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · United Kingdom | ||||||
| Roche Holding AG | Adopters | Switzerland | limited | solid | limited | limited |
Roche uses artificial intelligence mainly in diagnostic products such as digital pathology and the assessment of kidney risk as well as in research, and has adopted its own ethical principles for this. On security there is an external certification of information security, a public channel for reporting vulnerabilities and regulatory reviews of individual products, while board oversight of AI and precise information on the data and limits of the AI models were not found. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · Switzerland | ||||||
| S&P Global | Adopters | USA | limited | solid | limited | limited |
S&P Global has anchored oversight of technology and cyber risk in the board's audit committee, trained its employees group-wide in the use of artificial intelligence and describes its information protection with external tests and audits. However, public principles on the use of artificial intelligence, test reports on the reliability of its AI products and precise information on the models used are missing. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Salesforce | Adopters | USA | solid | solid | solid | solid |
Salesforce has anchored oversight of artificial intelligence in a board committee, reviews new AI applications before deployment and has had its AI management system certified under an international standard. Security researchers repeatedly found flaws through which AI agents could disclose customer data; Salesforce fixed them, and several court cases over data theft and training data are still pending. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %solid Index assessment: solid Role · CountryAdopters · USA | ||||||
| Samsung Electro-Mechanics | Enablers | South Korea | limited | solid | limited | limited |
Samsung Electro-Mechanics makes components for AI data centres and uses its own AI mainly for quality inspection in manufacturing; it publishes an AI policy, but responsibilities, review rules and controls for its use of AI are not described in a traceable way. The company documents the protection of its data and systems with recognised security certificates for several plants, and no court cases, penalties or security incidents were found in the period examined. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · South Korea | ||||||
| Samsung Electronics | Enablers | South Korea | solid | solid | limited | solid |
Samsung has set rules and responsibilities for artificial intelligence, says it tests its models before deployment and labels images edited with artificial intelligence, but discloses hardly any test results or information on its models. Over the collection of viewing habits via smart TVs, Samsung reached an agreement with the state of Texas on a consent requirement, and further lawsuits on the matter are pending. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryEnablers · South Korea | ||||||
| SanDisk Corp | Enablers | USA | limited | solid | limited | limited |
Sandisk makes memory for AI data centres and protects its systems and data with a documented security programme, a vulnerability reporting office and a data protection contact. How the company steers and monitors its own use of AI it has so far disclosed hardly at all. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Sanofi SA | Adopters | France | limited | solid | limited | limited |
Sanofi uses artificial intelligence widely in research, trials, drug safety and production and has principles, an internal review procedure for high-risk applications and a cybersecurity organisation described in its annual report. Who currently oversees these rules at board or executive committee level, and how well individual systems are tested, the company does not disclose. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · France | ||||||
| Schneider Electric SE | Enablers | France | limited | solid | limited | limited |
So far, Schneider Electric describes its own bodies and rules for the responsible use of artificial intelligence only in blog posts and interviews; what is documented is mainly an agreement with the trade unions in France under which employee representatives review new AI projects in advance. The security of product development is partly independently certified, but in 2024 the company was hit by two cyberattacks on its own systems and discloses little about the limits and data use of its AI products. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · France | ||||||
| Seagate Technology | Enablers | Ireland | limited | solid | limited | limited |
Seagate makes hard drives that are in demand in data centres for artificial intelligence and itself uses learning software in manufacturing, for example for defect detection, but discloses little about the rules and controls that apply to it. For the protection of its IT systems, the company describes oversight by a board committee as well as regular reviews and exercises, and no material cyber incident was reported during the review period. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Ireland | ||||||
| Siemens AG | Enablers | Germany | limited | solid | limited | limited |
Siemens uses artificial intelligence widely in software, assistants and controls for industry and buildings; the supervisory board has dealt explicitly with AI and cybersecurity, and one executive board member is responsible for data and AI. Protection against cyberattacks and the handling of security vulnerabilities are well documented; however, Siemens does not publish its own rules and approval procedures for the use of AI, and test results for the AI products and information on the models used are missing. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Germany | ||||||
| Siemens Energy | Enablers | Germany | limited | solid | limited | limited |
Siemens Energy has created a dedicated committee for digitalisation and artificial intelligence on its supervisory board and describes its own cybersecurity organisation with a reporting point for vulnerabilities, but publishes no rules or review procedures of its own for the use of AI. The company has not yet publicly explained how the AI in its power plant and grid products is tested and where its limits lie. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Germany | ||||||
| SK hynix | Enablers | South Korea | limited | limited | limited | limited |
SK hynix increasingly uses artificial intelligence in its own administration and manufacturing and describes this mainly in press releases, without publicly available rules, responsibilities or reviews for this use. Independent researchers have also demonstrated a security flaw in the company's memory modules, and a former employee was convicted for taking manufacturing know-how. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · South Korea | ||||||
| SpaceX | USA | limited | limited | solid | limited | |
SpaceX, which since February 2026 has included the AI developer xAI with the chatbot Grok and the platform X, publishes technical descriptions with safety tests for its AI models and has set rules against misuse. However, the Canadian privacy regulator found that a Grok image tool was launched without adequate safeguards and enabled sexualised images of real people without their consent; further regulatory proceedings are under way in Europe and the US. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %solid Index assessment: limited Role · CountryInnovator · USA | ||||||
| Standard Chartered PLC | Adopters | United Kingdom | limited | solid | limited | limited |
Standard Chartered uses artificial intelligence widely, for example as an internal assistant for employees, in fighting financial crime and in client advice, and describes a body that reviews AI applications before use as well as extensive AI training. A body explicitly responsible for AI on the board or in executive management, published AI rules and explanations for customers are largely missing, and the supervisors in Hong Kong and Singapore have requested explanations after the announced job cuts that the bank attributes to AI. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · United Kingdom | ||||||
| Symbotic Inc | USA | limited | solid | limited | limited | |
Symbotic builds learning warehouse robots, mainly for Walmart, and describes its protective measures against cyber attacks and the ongoing monitoring of its systems, but does not disclose its own rules for dealing with artificial intelligence. In 2026 the company had to recall and replace certain components of its systems, and media reports describe frequent robot failures at its main customer. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryInnovator · USA | ||||||
| Synopsys Inc | Enablers | USA | solid | solid | limited | solid |
Synopsys has the risks of artificial intelligence overseen by its board, has new uses of AI assessed by an internal body and protects its systems with audited security procedures, but publishes neither the wording of its AI policy nor much detail on which models its tools use and where their limitations lie. A pending investor lawsuit and unauthorised access to email accounts containing personal data in 2025 remain unresolved. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryEnablers · USA | ||||||
| Taiwan Semiconductor Manufacturing Company | Enablers | Taiwan | limited | solid | limited | limited |
TSMC protects its fabs and data with an extensive security organisation, international standards and regular attack exercises, and its own controls uncovered a leak of trade secrets that was punished in court. The company only published rules for the responsible use of artificial intelligence in summer 2026, their implementation is not yet documented, and in the US an investigation is under way because chips made by TSMC ended up in AI processors of a sanctioned group. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Taiwan | ||||||
| Tencent Holdings | Adopters | China | solid | solid | limited | solid |
In its sustainability report, Tencent describes concrete responsibilities and review steps for artificial intelligence, such as safety reviews before launch, attack tests and the labelling of AI-generated content, and has individual units certified. Independent assessors, however, criticise that Tencent discloses little about how its recommendation systems and models work and which data they use, and its AI assistant Yuanbao drew attention twice in 2026 with insulting answers. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · China | ||||||
| Teradyne | Enablers | USA | limited | solid | limited | limited |
Teradyne has explicitly assigned oversight of artificial intelligence to the board of directors and two of its committees, but has not published binding rules of its own for dealing with artificial intelligence. The company demonstrably fixes security vulnerabilities in its robots in orderly procedures, most recently for a critical vulnerability in May 2026, but discloses very little about how the learning functions of the robots are tested and what happens to the data collected in the process. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Tesla Inc | Adopters | USA | limited | limited | limited | limited |
Tesla has set out board oversight of artificial intelligence but publishes no rules of its own for its use, and it has received a European approval for its supervised driving system. However, a court ruling on a fatal crash, several ongoing investigations by the US road safety authority and a Californian decision on misleading advertising show considerable weaknesses in safety and openness. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Travelers Companies | Adopters | USA | limited | solid | limited | limited |
Travelers uses artificial intelligence widely in claims handling, risk assessment and customer dialogue; the board deals with the topic and data security is described in detail, after New York authorities fined shortcomings in an older data breach in 2024. The company does not publish its own rules and approval procedures for artificial intelligence, and discloses little about how the systems are tested and whether customers can tell when they are talking to a machine. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| UBS Group AG | Adopters | Switzerland | solid | solid | limited | solid |
UBS uses artificial intelligence widely, for example as an internal assistant for most of its employees, in investment advice and in screening suspicious payments, and describes its own AI leadership, an internal AI policy, mandatory training and an extensive programme to protect against cyber attacks. So far the bank discloses little about how clients are informed about the use of AI and how the systems are tested; in 2025 employee data reached the internet via a supplier. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · Switzerland | ||||||
| Unimicron Technology Corp | Enablers | Taiwan | limited | limited | limited | limited |
Unimicron makes carrier substrates and circuit boards, a large share of which are built into computers for artificial intelligence, but discloses little about how it uses AI itself and who is responsible for it. In early 2025 a Chinese subsidiary was hit by an extortion attack with a suspected leak of internal data, and since August 2026 prosecutors have been investigating suspected false statements of origin, without any ruling so far. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · Taiwan | ||||||
| Union Pacific Corp | Adopters | USA | limited | solid | limited | limited |
Union Pacific uses artificial intelligence mainly to plan rail operations and to inspect track, and has assigned oversight of these risks to a committee of its board and to an internal expert body, but does not disclose how new AI applications are reviewed and approved. More detailed information on testing and the limits of the systems is missing, and two court cases concerned an automated locomotive control system and the collection of truck drivers' fingerprints. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| United Parcel Service | Adopters | USA | limited | solid | limited | limited |
UPS uses artificial intelligence widely, for example in customer service, customs clearance and in assessing the risk of theft at delivery addresses, and in 2025 it assigned oversight of these risks to a committee of its board. The company has not published its own principles, explanations of individual systems or a report on its use of AI, and in Sweden employees have reported a camera system in delivery vehicles to the data protection authority. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Vertiv Holdings | Enablers | USA | limited | solid | limited | limited |
Vertiv supplies power and cooling technology for artificial intelligence data centres and has offered an AI-assisted maintenance service since 2026, but discloses no rules of its own for the use of AI and no testing information on this service. To protect its systems the company describes oversight by the board, regular security testing and an independent certification, and reported vulnerabilities in devices have been fixed. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · USA | ||||||
| Visa Inc | Adopters | USA | solid | solid | limited | solid |
Visa uses artificial intelligence mainly to detect fraud in the card network and is building secure payments by digital shopping assistants; the board has explicitly assigned oversight of the risks of this technology to a committee, and employees are trained in its responsible use. Public descriptions of the individual systems and their limits, as well as an independent review, are still missing, and authorities and shareholders are calling for more protection against money being made over the network with fake intimate images. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · USA | ||||||
| Walmart Inc | Adopters | USA | solid | solid | limited | solid |
Walmart has explicitly assigned oversight of artificial intelligence to board committees and describes rules for privacy and security, but publishes little on how individual systems are tested and explained. Several lawsuits accuse Walmart of collecting biometric data without adequate consent in the facial verification of delivery drivers and in an automated phone system; one lawsuit ended in a settlement, and no court has yet ruled on the others. Responsibility · 40 %solid Safety · 40 %solid Transparency · 20 %limited Index assessment: solid Role · CountryAdopters · USA | ||||||
| Wells Fargo & Co | Adopters | USA | limited | solid | limited | limited |
Wells Fargo has anchored oversight of artificial intelligence at board level and appointed a person responsible for AI, but discloses no binding AI policy and no documented procedure for reviewing and approving AI applications. The company describes security precautions, for example that customer data is removed before use by the digital assistant; publicly available explanations of individual AI systems are largely missing, and a lawsuit over possible discrimination in automated mortgage decisions is still pending without a court finding. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · USA | ||||||
| Willis Towers Watson | Adopters | United Kingdom | limited | solid | limited | limited |
WTW has assigned oversight of risks from artificial intelligence to a board committee and describes fixed procedures for handling security incidents in its annual report, but does not publish its internal rules for the use of AI. How the AI tools it increasingly uses for policyholders and plan participants are tested and disclosed to those affected is something the company has so far barely explained. Responsibility · 40 %limited Safety · 40 %solid Transparency · 20 %limited Index assessment: limited Role · CountryAdopters · United Kingdom | ||||||
| Z.AI Co Ltd | China | limited | limited | solid | limited | |
Z.ai discloses bodies for ethics and data security and publishes its language models together with technical reports, but had not published the safety framework it promised in 2024 by mid-2026. A US agency rates its latest open model as the strongest open model to date for cyberattacks, and since 2025 the US government has kept the company on an export blacklist because of military links, which Z.ai disputes. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %solid Index assessment: limited Role · CountryInnovator · China | ||||||
| Zhongji Innolight Co Ltd | Enablers | China | limited | limited | limited | limited |
Zhongji Innolight is one of the largest manufacturers of optical transceiver modules for artificial intelligence data centres, but discloses hardly anything about how it uses artificial intelligence in its own manufacturing, who is responsible for it and how it is controlled. Since June 2026 the US Department of Defense has listed the company among firms with alleged links to the Chinese military, which the company disputes; no cases of harm caused by artificial intelligence were found. Responsibility · 40 %limited Safety · 40 %limited Transparency · 20 %limited Index assessment: limited Role · CountryEnablers · China | ||||||
No company found.
Click a column header to sort, also by each pillar.
How we assess.
Everything is open to scrutiny, but folded away. If you only want the result, you can skip the method.
The method in 50 secondsResearch, three pillars, the Anthropic example, human approval.Why this question is becoming more important
362 AI incidents in 2025 · 2027 Obligations in the EU · 46 % trust AI
362 incidents in which AI caused or nearly caused harm were documented worldwide in 2025. That is 55% more than the previous year and more than ever recorded. When AI helps decide on loans, job applications or diagnoses, a single error affects not one person but millions.
Quelle: Stanford HAI ↗From December 2027, EU obligations for high-risk AI applications — including credit decisions and personnel selection — take effect. Companies that are prepared will pay less to comply.
Quelle: EU AI Act ↗Only 46% of people worldwide are willing to trust AI, even though two thirds already use it regularly. Some 70% consider regulation necessary. Companies that disclose their practices gain trust.
Quelle: KPMG ↗Four levels
strong 3, solid 2, limited 1, unclear 0 points; harm rule where harm is documented.
Each pillar receives one of four ratings: strong 3 points, solid 2, limited 1, unclear 0. Weighted at 40, 40 and 20 per cent, they produce the overall rating: from 2.5 points “strong”, from 1.6 “solid”, provided no pillar is “unclear”, from 0.9 “limited”. Two “solid” pillars and one “limited” therefore give “solid”. Harm rule: if a court ruling, regulatory decision, fine or pattern within the last 36 months documents harm from the use of AI, the overall rating cannot be better than “limited” as long as no remedy is documented.
Published, specific, verifiable evidence and no moderate or serious controversy.
Substantiated structures, but sometimes lacking specificity or relevance to the product.
Only principles or declarations of intent, or the harm rule applies.
No disclosure found. That is a transparency gap, not misconduct. Whoever does not disclose cannot show responsible conduct.
What the assessment means
A rating, not an exclusion from the index. Not a certification or quality label.
The Globalance AI Footprint is an assessment. It is published for every company but has no influence on inclusion, ranking or weight in the GI 100 Index. In the GI Economy Portfolio, companies to which the harm rule applies or with a documented weakness may be excluded. The rating is not a certification or quality label.
All rules in detail: pillars, review questions, evidence levels, overall rating, harm rule and objections.
How Globalance itself uses AI.
The Globalance AI Footprint asks of every company how artificial intelligence is changing its business. The same question applies to Globalance. Three places where AI already contributes today:
Research and measurement
Globalance Intelligence uses AI to evaluate more than 440 sources every week. Every figure in the index has an open source, and a nightly check compares the values with the sources.
Analysis
The Globalance AI Footprint of every stock in the GI 100 Index is produced with AI support and is checked and approved by Globalance analysts.
Decision
Investment decisions are made by people. AI provides proposals; it does not decide.