Greenwashing: How AI can substantiate green claims

As businesses increasingly highlight their sustainability efforts, the risk of greenwashing has grown.

Rob Biddlecombe, Partner at Mills & Reeve, shares his insights on the legal framework governing green claims, how AI can substantiate these claims, and the implications for businesses.

The law on green claims

Across all types of advertising, claims must be substantiated by evidence and must not mislead consumers.

Adverts that include “green claims” on sustainability and environmental impact are particularly under scrutiny currently.

If companies claim to be more eco-friendly and sustainable than they actually are, they are at risk of greenwashing.

Advertising is governed by the advertising codes (BCAP for broadcast adverts; CAP for non-broadcast) and the Competition and Markets Authority’s (CMA) Green Claims Code.

The Advertising Standards Authority (ASA) and CMA assess whether advertisers are complying with codes and can enforce regulations, including through carrying out investigations.

For example, they recently investigated fast fashion brands in relation to how they market their green credentials.

Both the ASA and CMA have a keen ongoing focus on tackling greenwashing as it has been identified a policy priority because of the potential for harm to consumers.

A new AI chatbot

A new Chinese AI startup has recently made headlines as it launched DeepSeek “R1”, built on DeepSeek “V3” which was released in December 2024.

The DeepSeek app quickly topped app store charts in the US, causing some alarm among US-based tech companies.

The app is powered by DeepSeek V3 and has been compared to OpenAI’s o1 model in terms of its abilities. The chatbots appear to be similar in style, capability, and the types of tasks they are used for.

There’s been lots of discussion on the cost effectiveness of DeepSeek. DeepSeek V3 has reportedly been produced at a cost of around $6 million, in contrast to potentially over $100 million spent on GPT-4.

Besides the cost, DeepSeek is claimed to be less harmful to the planet.

This is an important claim as the environmental impact of AI generally is significant, with carbon dioxide emissions for ChatGPT estimated at over 260 tonnes per month.

DeepSeek claims to be a greener alternative to ChatGPT because it requires much less computational power and electricity so generates less carbon dioxide emissions.

It seems to require fewer computer chips than other AI platforms, but still produce similar output. Despite this, there are potential opportunities for AI to contribute towards addressing the climate crisis.

For example, AI can contribute to tracking weather and pollution. It’s been suggested that AI can be used to track changes in icebergs and map the impact of deforestation far more efficiently than humans could.

It is thought that AI could even be used to help with predicting climate disasters which could in turn enable organisations to mitigate the impact of disasters.

On the topic of greenwashing, AI can be used positively as a tool to identify greenwashing.

It can analyse reports and sustainability claims and compare them to data on environmental impact, and it can do this faster than humans could.

AI and green claims

Although it appears that AI chatbots are becoming more efficient, there are still lots of questions to be answered before any AI model or any business that uses an AI model would be able to make a green claim.

Generally speaking, only businesses and products that have no overall negative impact on the environment can make absolute claims that they are "green" or "sustainable".

Therefore, even if an AI model is able to demonstrate that it uses less computational power and chips than other AI, it still likely cannot make an absolute claim because it still uses significant computational power and chips.

That is unless the specific use of the AI has such a significant positive impact on the environment that the environmental impact of the computational power and chips is offset.

This means AI models, or any business using AI models, would in most case only be able to make a more nuanced claim perhaps about how they are greener or more sustainable than other AI models – or maybe that they are on a mission to become more sustainable over time, by reducing their carbon footprint.

Whichever claim is made, businesses which rely on AI models will need robust data to demonstrate that their claim is true across the entire life cycle of the development training rollout and use of the AI model.

This is required as a matter of law but is particularly important in the context of claims which are likely to be highly scrutinised and challenged by competitors using other AI models.

How we can help you

Greenwashing remains a hot topic, and the ASA and CMA are enforcing rules on misleading advertising in relation to green claims.

Our specialist team are experienced in assisting organisations with the substantiation of environmental claims and responding to ASA and CMA investigations.

Please get in touch with our experts Katrina Anderson, Jessica Mason, or Rob Biddlecombe, if you would like to discuss your claims on sustainability or find out more about regulations on greenwashing.

Pictured: Rob Biddlecombe