Using AI? Check Your Environmental Advertising, Including ESG Reports and SEC Filings
by: Rosanne Yang and Sophia Allen
The speed of AI development and adoption brings promises of better, faster, and potentially cheaper results. It puts vast computing power in the hands of non-tech business teams, enhancing and broadening access to businesses of all shapes and sizes. All of this is great news for businesses and the customers they serve.
However, as with most things, the use of AI should be examined from many angles, one of which is its environmental impact. While we are still in the early days of understanding current and projected environmental impacts of AI usage, the available evidence points to those impacts being significant – so much so that some have labeled AI as incompatible with notions of sustainability.
Companies should consider these impacts before adopting AI tools to begin with. And then when a company takes the affirmative step of putting out an ESG report, making a statement in its SEC filing about the company’s efforts around sustainability, or putting environmental claims into its product or service advertising, it must take care to properly account for its use of AI in those statements to ensure they are not misleading, deceptive, or straight out false.
What are these environmental impacts and how do they interplay with your advertising?
1. Carbon Emissions and Energy Consumption
As a refresher, greenhouse gas emissions are generally categorized into Scope 1, 2, and 3 emissions:
Scope 1 emissions are direct emissions from your owned and operated operations, such as the fuel burning in your office’s boiler.
Scope 2 emissions are indirect emissions, generated by the energy sources that you purchase for your operations, such the emissions that are generated from the power plant that produces the electricity you buy.
Scope 3 emissions are indirect emissions, beyond those in Scope 2 that occur in the upstream (e.g., from suppliers) and downstream (e.g., from consumers) activities of your value chain. Examples range greatly, from employee commuting and business travel to your purchased goods and services to the transportation of your own goods to your consumer by a third party. Scope 3 emissions can account for a significant portion of your emissions.
The Scope 3 bucket is also likely where a lot of companies’ AI-related emissions will stack up. And they will stack up due to AI’s appetite for energy. For example, it has been reported that, “One query to ChatGPT uses approximately as much electricity as could light one lightbulb for about 20 minutes.” And, when considering that your own employees’ use of ChatGPT on the job could contribute to your emissions, this can quickly become a wildfire scenario.
Carbon claims are increasingly being challenged by consumers and specifically regulated, including by the FTC and the state of California. If you are making any carbon claims, especially aspirational claims regarding reductions in your carbon emissions or carbon neutrality, take special care to understand how your (and your suppliers’) AI-use contributes to your carbon footprint.
2. Water Usage in Chip Making and Cooling
Data centers require vast amounts of water to keep them cool enough to operate. Microsoft’s 2022 Envrionmental Sustainability Report disclosed that it had a 34% rise in its water consumption, which has been attributed by many to Microsoft’s AI operations. It has also been reported that a single ChatGPT conversation uses approximately 17 fluid ounces of water.
Chip making also requires vast amounts of water. The World Economic Forum reports that the average chip manufacturing plant can use 10 million gallons of ultrapure water per day – the equivalent of the water usage of 33,000 U.S. households.
If your advertising materials, ESG statements, and/or SEC filings discuss your company’s water use reductions or reclamation be sure that those calculations and statements incorporate and reflect the water usage arising from your company’s uses of AI.
3. Waste
Chips and other hardware associated with the computing behind AI do not last forever and may need to be replaced every few years to keep up with fast-developing innovations. Some chip makers are promising new chip versions every year as a result of the speed of development. While it may be possible to recycle or reuse some of these materials, it is undeniable that the amount of e-waste is burgeoning. The World Economic Forum projects that more than 120 million metric tons of e-waste will have been generated by 2050. In addition to sheer quantity of waste that is often added to landfills, e-waste contains a variety of hazardous chemicals which can leech into the soil and water supplies. Additionally, waste can contribute to Scope 3 emissions.
If your public statements include waste reduction factors, ensure that they account for the added e-waste associated with AI adoption and/or are qualified to avoid misleading readers as to the scope and nature of the reductions.
In summary, companies who release ESG reports or make other sustainability claims with respect to their products or operations, directly or in their supply chains, should re-evaluate those statements in light of their or their suppliers’ use of AI to ensure that they remain accurate or are further qualified or revised to account for the increased emissions, water usage, and waste that are part and parcel of AI use.
Originally published by InfoLawGroup LLP. If you would like to receive regular emails from us, in which we share updates and our take on current legal news, please subscribe to InfoLawGroup’s Insights HERE.