The Dark Side of Corporate AI Hype: More Tech Talk, Less Environmental Action

Artificial intelligence promises cleaner production and greater industrial efficiency, but research reveals a troubling divide between companies promoting AI and those investing in it. The findings expose how technology-driven market expectations can undermine environmental spending, weaken transparency and create new challenges for corporate sustainability governance.

A study titled "The Environmental Cost of AI Hype: Corporate AI Catering, Pollution Emissions, and Sustainability Governance," published in the journal Sustainability, examines this relationship. Conducted by Jingwei Hu of Shanxi University of Finance and Economics and Shuaiwen Chang of Central South University, the research investigates whether corporate AI promotion can influence environmental performance through investment decisions, disclosure practices and market incentives.

Analyzing 2,784 Chinese listed companies between 2016 and 2024, the researchers identify a significant association between excessive AI-related disclosures and higher pollution intensity. Their findings also reveal a contrasting relationship: substantive AI investment is associated with lower pollution, suggesting that environmental risks arise not from technological adoption itself but from the gap between corporate promises and measurable action.

When AI Promotion Becomes More Valuable Than AI Investment

The rapid expansion of AI has created powerful incentives for companies to present themselves as technologically advanced. Investors increasingly recognize the potential of AI to improve productivity, strengthen competitiveness and transform industrial operations. However, developing these capabilities requires sustained expenditure, specialized expertise and organizational restructuring.

Corporate managers consequently face competing pressures. Genuine technological transformation involves uncertain outcomes and long implementation periods, while favorable market reactions to AI-related announcements may arrive much sooner. The research examines how this imbalance can encourage companies to emphasize technological progress without making corresponding investments.

The authors describe this behavior as AI catering, referring to AI-related disclosures that exceed the level justified by observable substantive technological input. Unlike straightforward technology adoption, AI catering reflects a discrepancy between the technological image a company communicates and the underlying investment supporting that image.

The distinction also separates AI catering from conventional greenwashing. Greenwashing concerns misleading environmental claims, whereas AI catering involves exaggerated technological narratives whose environmental consequences may emerge indirectly. Companies do not necessarily need to make questionable sustainability promises for their technology-focused behavior to coincide with weaker pollution control.

To investigate this relationship, the researchers examined 15,807 firm-year observations covering Chinese A-share listed companies. They analyzed AI-related language in annual reports using 73 keywords and compared disclosure patterns with AI patent applications, research expenditure and company characteristics.

Pollution intensity was assessed using industrial emissions of sulfur dioxide, nitrogen oxides, chemical oxygen demand and ammonia nitrogen, adjusted for operating revenue. This approach enabled the researchers to examine whether differences between technological narratives and observable AI investment corresponded with differences in environmental performance.

The analysis found that corporate AI disclosure expanded considerably faster than substantive AI input over the period examined. The widening discrepancy was concentrated among a subset of companies, indicating that technological promotion and measurable investment did not advance uniformly across the corporate sector.

The Pollution Divide Between AI Claims and Real Technological Investment

The study's central finding challenges the assumption that greater corporate attention to artificial intelligence necessarily signals environmental progress. Companies exhibiting higher levels of AI catering also demonstrated significantly higher pollution emission intensity, even after accounting for numerous financial and organizational characteristics.

A one-standard-deviation increase in excessive AI disclosure corresponded to approximately 4.4% higher pollution intensity. Moving from the 25th to the 75th percentile of AI catering was associated with approximately 1.9% higher intensity, establishing a measurable relationship between technological promotion and environmental performance.

The contrast became particularly revealing when the researchers examined AI disclosure and substantive investment separately. A one-standard-deviation increase in nominal AI disclosure was associated with approximately 4.2% higher pollution intensity, holding substantive input constant. An equivalent increase in substantive AI input, however, corresponded to approximately 3.2% lower intensity.

These opposing relationships suggest that the environmental significance of corporate AI activity depends partly on whether companies are developing technological capabilities or primarily communicating them. Observable investment was associated with better pollution outcomes, while promotional disclosure beyond expected investment levels showed the opposite relationship.

The researchers investigated potential explanations through environmental investment, disclosure quality and standardized environmental management. Their analysis identified two statistically supported channels through which AI catering may be connected to weaker environmental performance.

Companies with greater excessive AI disclosure tended to allocate less investment toward environmental protection. Such expenditure includes wastewater treatment, pollution-control equipment, emissions treatment and other activities requiring sustained financial commitments.

Environmental disclosure quality also deteriorated as AI catering increased. Companies exhibiting stronger promotional behavior provided less comprehensive environmental information, potentially weakening the ability of investors, regulators and other stakeholders to assess pollution-related risks.

Both relationships are consistent with the possibility that short-term technology-focused priorities compete with environmental commitments for corporate resources and managerial attention. The analysis nevertheless treats these mechanisms as correlational evidence rather than definitive proof of how AI catering causes pollution.

A third proposed channel involving standardized environmental management received no statistically significant support. Using ISO 14001 certification as an indicator, the researchers could not confirm that excessive AI promotion weakened formal environmental management systems. Certification status, however, does not establish how effectively such systems operate.

Why Manufacturers Face Greater Risks—and Scrutiny Changes the Outcome

The relationship between AI catering and pollution was not equally strong across different business environments. Traditional manufacturing companies exhibited a stronger positive association, highlighting how industrial characteristics can influence the environmental consequences of technology promotion.

These businesses often depend on physical equipment, established production processes and specialized workforce capabilities. Integrating AI into existing operations may require substantial machinery upgrades, redesigned processes and organizational adjustments, increasing the cost and complexity of meaningful adoption.

The researchers found that traditional manufacturers discussed AI at approximately the same level as other companies while demonstrating lower substantive AI input. Their wider disclosure-investment gap was consistent with the greater difficulties associated with transforming established industrial operations.

Pollution control in these industries also relies heavily on continuing environmental expenditure. Consequently, corporate decisions that coincide with reduced environmental investment may be especially consequential for businesses operating pollution-intensive production systems.

External oversight appeared to moderate these relationships. Companies receiving greater financial analyst coverage exhibited a weaker association between excessive AI disclosure and pollution intensity, suggesting that closer scrutiny can constrain the environmental consequences accompanying promotional behavior.

Environmental enforcement produced a similar pattern. Exposure to China's central environmental inspections was associated with a significantly weaker relationship between AI catering and pollution, consistent with stronger regulatory deterrence limiting reductions in pollution-control efforts.

Neither analyst coverage nor inspection exposure was significantly associated with the overall level of AI catering. This distinction suggests that external scrutiny may restrict its environmental consequences without necessarily eliminating the promotional behavior itself.

The research also examined whether financial markets provide incentives for exaggerated AI narratives. Companies exhibiting higher AI catering tended to have higher market valuations and greater ownership by short-term institutional investors, alongside lower ownership by long-term institutional investors.

Such patterns are consistent with a market environment in which short-term expectations can reward technological promotion. They do not establish that promotional disclosures directly cause valuation gains, but they reinforce the relevance of financial incentives when assessing corporate sustainability decisions.

Closing the Gap Between Technological Ambition and Environmental Accountability

The findings carry significant relevance for corporate disclosure regulation, investment assessment and environmental governance. They indicate that evaluating AI adoption through promotional language alone risks confusing technological ambition with meaningful operational transformation.

Regulators could address this problem by requiring companies to substantiate AI-related statements through verifiable evidence, including research expenditure, procurement contracts and identifiable applications. Stock exchanges could also scrutinize substantial discrepancies between technological claims and observable investment.

Environmental reporting requires parallel attention. More comprehensive disclosure of pollution emissions and environmental capital expenditure could help external stakeholders assess whether companies pursuing technological modernization continue to meet their environmental responsibilities.

For investors, the research suggests examining technological capabilities and pollution outcomes together rather than treating AI narratives as independent indicators of corporate progress. Such assessments could provide stronger grounds for distinguishing genuine innovation from short-term reputation management.

The researchers nevertheless acknowledge important limitations. Patent applications cannot capture AI acquired through external suppliers, licensed models or other arrangements, potentially overstating the gap between disclosure and actual adoption.

Environmental investment information was also available for only part of the sample, while the evidence concerns Chinese listed companies operating under particular market and regulatory conditions. The statistical findings therefore cannot automatically be generalized to unlisted businesses or other countries.

Although alternative statistical models and additional tests reinforce the central association, they do not conclusively establish causality. Further research incorporating AI procurement, operational implementation and cross-country comparisons would help clarify the underlying relationships.

The broader lesson concerns the governance of technological transformation. AI can contribute to cleaner production, but companies pursuing recognition for technological progress must also maintain credible environmental commitments. Without verifiable investment, reliable disclosure and effective oversight, enthusiasm for innovation may coexist with deteriorating pollution performance rather than deliver the environmental improvements it promises.