AI Ethics Scorecard: Nvidia, Meta, and Google
You own these companies. If you hold an index fund, a tech ETF, or a retirement account with equity exposure, Nvidia, Meta, and Alphabet sit in your portfolio. Together with Microsoft, Amazon, and Apple, they account for a commanding share of the S&P 500 by market capitalisation.
The question is not whether you are invested in AI. You are. The question is whether you know what you own.
Mashinii scored the major AI companies across 11 ethical dimensions using court filings, regulatory actions, investigative journalism, and NGO reports. No corporate self-assessments. No ESG questionnaires. Every score is backed by cited, independently sourced evidence processed through our scoring methodology.
The results should concern any investor paying attention.
How the Major AI Companies Compare
Our Safe & Smart Tech dimension is where AI ethics sits. It measures data breach severity, AI governance, cybersecurity investment, privacy certifications, algorithmic transparency, user data control, and vulnerability management across 19 individual KPIs.
But technology safety alone does not capture the full picture. A company building frontier AI while supplying military systems, facing antitrust proceedings, or accumulating data privacy fines raises compounding ethical questions. The full 11-dimension view matters.
| Company | Ticker | Safe & Smart Tech | Average Score |
|---|---|---|---|
| Salesforce | CRM | +30 | +11.8 |
| Apple | AAPL | +20 | +9.5 |
| Microsoft | MSFT | +10 | +5.5 |
| Alphabet/Google | GOOGL | -30 | +2.3 |
| Amazon | AMZN | -30 | -10.0 |
| Meta Platforms | META | -50 | -14.5 |
| Oracle | ORCL | -50 | -13.6 |
NVIDIA (Avg: -9.1) is covered separately below. Its primary ethical concerns in our data relate to defence AI applications and supply chain practices rather than consumer data privacy.
Three patterns are immediately visible. The companies dominating AI headlines are not the ones scoring well on technology safety. There is an 80-point spread between the best and worst performers. And only three of these seven companies scored positively on data privacy and AI governance at all.
Nvidia: The Hardware Layer
Average Score: -9.1
Nvidia is the most valuable semiconductor company in history. Its GPUs underpin virtually every major AI training run. From a technology standpoint, its position is unassailable.
Ethical integrity, however, is not measured by market capitalisation. Mashinii scored Nvidia at -9.1 across 11 dimensions. The two lowest scores tell a specific story.
| Dimension | Score |
|---|---|
| No War, No Weapons | -40 |
| Fair Trade & Ethical Sourcing | -30 |
The -40 on weapons reflects documented involvement in defence AI applications identified in public procurement records, according to our analysis. Nvidia's chips power military AI systems, surveillance infrastructure, and intelligence-gathering platforms across multiple countries. For a company whose technology is foundational to AI, who uses that technology is not a peripheral question.
The -30 on ethical sourcing relates to supply chain concerns inherent to semiconductor manufacturing: rare earth mineral extraction, fabrication facility labour practices, and the concentrated geography of chip production. These challenges span the industry, but they are documented in Nvidia's supply chain specifically.
An overall average of -9.1 places Nvidia in the bottom half of S&P 500 technology companies in our ranking. Traditional ESG frameworks, focused on corporate disclosure quality, tend to underweight the downstream application risks that define Nvidia's ethical profile. We explored this gap in detail in our analysis of why ESG ratings miss the full picture.
View Nvidia's full score breakdown ->
Meta: The Model Builder
Average Score: -14.5
Meta is investing tens of billions in AI infrastructure. Its Llama models are among the most widely deployed open-source AI systems. The company positions itself as a leader in responsible AI development.
The independent record, according to our scoring, tells a different story. Meta scored -14.5 -- the lowest of the three title companies and 30th out of 40 major S&P 500 companies we analysed.
| Dimension | Score |
|---|---|
| Safe & Smart Tech | -50 |
| Honest & Fair Business | -40 |
| Fair Pay & Worker Respect | -30 |
A -50 on Safe & Smart Tech is among the lowest in our database for any major technology company on this dimension. According to our analysis, multiple jurisdictions have imposed fines and sanctions related to Meta's data handling practices. Regulatory actions documented in public records span data privacy violations, user consent issues, and the scope of personal data processing across Meta's platforms.
For a company building large-scale AI models trained on user data, that score carries particular weight. The scale of Meta's data collection -- across Facebook, Instagram, WhatsApp, and its AI training pipelines -- means governance failures compound at an unusual rate.
The -40 on governance reflects regulatory penalties and antitrust proceedings extending well beyond data privacy. The -30 on worker respect is driven by documented concerns about content moderation contractor conditions identified in public reporting. Our data privacy deep dive covers Meta's position relative to the worst-scoring companies in our database.
View Meta's full score breakdown ->
Alphabet/Google: The Search and Model Giant
Average Score: +2.3
Google is the third-largest investor in AI infrastructure globally. DeepMind is a pioneer in frontier research. Google Search, Gmail, Android, Chrome, and YouTube reach billions of users daily. The company has more data on more people than arguably any organisation in history.
Mashinii scored Google at +2.3 overall. Barely positive. The dimension-by-dimension breakdown reveals sharp contradictions.
The -30 on Safe & Smart Tech reflects the fundamental tension in Google's business model. According to our analysis, Google has faced multiple significant regulatory actions related to data handling, including GDPR fines, antitrust investigations citing data market power, and documented concerns about location tracking scope and cross-service data integration. For a company building AI systems that process information at global scale, this tension between collection and protection is structural, not incidental.
The -40 on governance -- its lowest dimension -- is driven by ongoing antitrust proceedings across multiple jurisdictions. The U.S. Department of Justice case, European Commission competition fines, and advertising market investigations all contribute.
Google's +40 scores on environmental performance and waste reduction are real. But they do not address the data ethics questions most relevant to its AI operations. An investor focused on responsible AI should weight the privacy and governance scores accordingly. We covered this dynamic between Google, Microsoft, and Apple in our privacy comparison.
View Google's full score breakdown ->
The Rest of the AI Landscape
Salesforce: Safe & Smart Tech +30, Average +11.8
The highest-scoring major AI company on technology safety in our analysis. According to our data, its +60 on governance -- the best governance score among major tech companies we cover -- reflects transparency practices and regulatory compliance extending to its AI products. Einstein AI operates within a framework that has generated fewer regulatory concerns than its peers.
View Salesforce's full score breakdown ->
Apple: Safe & Smart Tech +20, Average +9.5
Apple's +20 reflects its positioning of privacy as a core product feature. According to our analysis, App Tracking Transparency, on-device processing, and end-to-end encryption all contribute positively. But Apple's AI ambitions are growing, and a supply chain labour score of -30 on Fair Pay & Worker Respect represents a significant concern elsewhere. Our smartphone supply chain analysis covers this in detail.
View Apple's full score breakdown ->
Microsoft: Safe & Smart Tech +10, Average +5.5
Mildly positive on technology safety. Our data indicates significant enterprise cybersecurity investment and a formal Responsible AI framework. The score is limited by documented concerns around consumer product data collection. More critically, Microsoft carries a -60 on No War, No Weapons -- the lowest defence score among AI companies we cover -- reflecting extensive military contracts in public procurement records. For context on how defence exposure intersects with AI ethics, see our defence contractors analysis.
View Microsoft's full score breakdown ->
Amazon: Safe & Smart Tech -30, Average -10.0
Amazon's -30 reflects documented concerns about worker surveillance practices, including a 32 million euro fine in France for GDPR violations related to employee productivity tracking, according to our analysis. Data handling questions from devices and services across its ecosystem compound the score. Our Amazon vs Walmart comparison provides additional context.
View Amazon's full score breakdown ->
Oracle: Safe & Smart Tech -50, Average -13.6
Oracle matches Meta for the lowest Safe & Smart Tech score among major AI companies. According to our data, regulatory findings documented in public proceedings drive this result. A -40 on worker respect, based on employment-related legal proceedings, compounds the concern. For a company handling enterprise data at scale and expanding into AI cloud services, these scores warrant close scrutiny.
View Oracle's full score breakdown ->
The Pattern: AI Leaders Are Not Ethics Leaders
The data reveals a structural pattern. The companies investing most aggressively in AI tend to score worse on data privacy and technology safety.
| Category | Avg Safe & Smart Tech | Avg Overall |
|---|---|---|
| AI Infrastructure (NVIDIA) | * | -9.1 |
| AI Platform Builders (Meta, Google) | -40.0 | -6.1 |
| AI Enterprise (Salesforce, Microsoft) | +20.0 | +8.7 |
| AI-Integrated Consumer (Apple, Amazon) | -5.0 | -0.3 |
| AI-Adjacent Enterprise (Oracle) | -50.0 | -13.6 |
*NVIDIA's Safe & Smart Tech score is not among its documented key dimensions in our analysis. Its primary ethical concerns relate to defence applications and supply chain practices.
The companies with the most data and the largest AI ambitions -- Meta and Google -- carry the deepest negative scores on responsible technology. Enterprise-focused companies -- Salesforce and Microsoft -- score better, potentially because B2B business models create different incentive structures around data stewardship.
This is not coincidental. Business models built on advertising revenue require extensive data collection. Collection at scale creates friction with privacy regulation. That friction generates the regulatory actions and court proceedings that drive negative scores in our methodology. We documented this pattern across the broader S&P 500 in our ethical integrity analysis.
AI Governance: Principles vs. Conduct
Our Safe & Smart Tech dimension includes KPIs for AI ethics governance, AI audit practices, algorithmic transparency, and algorithmic harm remediation. These are the metrics most directly relevant to responsible AI development.
Salesforce (+30), Apple (+20), and Microsoft (+10) have all established formal AI ethics review processes, published responsible AI principles, and submitted to varying degrees of independent oversight. Whether those structures keep pace with the speed of capability advancement is an open question. But they exist, and they register in the independent record.
Meta (-50), Oracle (-50), Google (-30), and Amazon (-30) have governance programmes too. But our data shows a gap between stated principles and documented conduct. When a company publishes responsible AI principles while facing data privacy fines across multiple jurisdictions, our methodology weights regulatory actions over corporate statements.
Explore what Safe & Smart Tech measures ->
The Defence Dimension: AI Beyond Data Privacy
The use of AI in military applications is one of the most consequential ethical questions of the decade. Several AI companies carry significant scores on our No War, No Weapons dimension.
| Company | No War, No Weapons Score |
|---|---|
| Microsoft | -60 |
| NVIDIA | -40 |
| Amazon | -30 |
| -30 | |
| Meta | * |
| Apple | 0 |
*Not among Meta's documented key dimensions in our analysis.
Microsoft's -60 reflects the most extensive documented defence relationships among major AI companies. Nvidia's -40 reflects the role of its hardware in military AI systems. Amazon's -30 and Google's -30 both reflect documented defence and intelligence agency contracts, including AWS's share of the $9 billion JWCC cloud contract and Google's cloud and AI contracts identified in public procurement records.
For investors who screen on military involvement, these scores matter as much as data privacy. A company that handles user data responsibly but supplies AI-powered weapons systems presents a fundamentally different ethical question -- one that blended ESG ratings typically obscure. Our defence stocks analysis covers this dimension in depth.
What Blended ESG Ratings Obscure
Several companies in this analysis receive strong traditional ESG ratings. Nvidia has been a top holding in ESG-labelled technology funds, driven by energy-efficient chip design and disclosure quality.
Our data, scored across 11 independent dimensions, reveals trade-offs that blended ratings compress into a single number:
- Nvidia scores -40 on weapons involvement while receiving high marks from agencies that do not screen for military AI applications.
- Meta receives investment-grade ESG ratings while scoring -50 on data privacy -- the dimension most directly relevant to its AI operations.
- Google's +40 on environmental performance elevates its traditional ESG grade, while its -30 on data privacy and -40 on governance tell a different story.
Blended ratings allow strong environmental scores to mask poor data ethics. Mashinii does not blend. Each dimension stands independently, and investors weight them according to their own values. We wrote at length about this structural problem in why ESG ratings miss the full picture.
How Mashinii Scores Companies
Mashinii scores companies across 11 ethical dimensions using independently sourced data: court filings, regulatory actions, investigative journalism, and NGO reports. Scores range from -100 to +100. A score of 0 means insufficient independent evidence in either direction. Every score is backed by cited sources.
No corporate self-assessments. No ESG questionnaires. See the full methodology ->
What Does Your Portfolio Look Like?
If you hold AI exposure -- and most equity investors do -- the ethical profile of these companies is already part of your portfolio. The AI sector is concentrated. These six companies alone shape the ethical character of any index-tracking fund.
You do not have to guess. Mashinii shows you exactly how your holdings score across all 11 dimensions.
Or look up any company directly: