IS YOUR TECH PR AGENCY DOING THEIR DUE DILIGENCE ON AI?

You’re in a meeting with your tech PR agency explaining the details of your new technology. It’s a super-clever toaster that programmes your favourite browning setting for your sourdough or your white sliced bread. “That’s personalisation, that’s AI,” says the enthusiastic account manager. You consider this. It’s only two little letters but it’s a trend that media love and it will help get attention for your toaster tech. “Yes, let’s include it in the press release” you hear yourself saying.

Stop!

In 2024, several companies were hauled over the coals by media and regulators for making over-inflated claims about their use of AI, driven by PRs and marketers set on inserting their clients’ technology into this exciting new trend. In fact, it was so common, it became a phenomenon called “AI washing”.

Just like its predecessor, greenwashing, AI washing refers to companies who claim to use AI when they’re actually using simple algorithms for matching or personalisation. Or perhaps companies are using AI but only to drive one element of their service, an AI chatbot for customer service for example. Can these companies really claim that their technology is driven by AI when it’s just an add on?

It helps to understand what AI really means. While there is no exact definition, artificial intelligence relates to a software’s ability to understand and solve problems but critically also learn and apply that learning. It’s the learning that is necessary for the artificial to be intelligent, because the machine is only as good as what it has learned. Current AI models are being trained on huge amounts of information that is freely available on the internet, not all of it correct and not all of it free to use, but that is an AI trip hazard for another blog.

It’s a problem that regulators are noticing

In March 2024, the United States Securities and Exchange Commission (“SEC”) brought charges against two investment advisors for AI washing. The SEC Chair stated that these companies “marketed to their clients and prospective clients that they were using AI in certain ways, when in fact they were not”.

Last year, the BBC’s technology correspondent, Emma Woollacott wrote that AI washing had caught out mighty tech giant Amazon. Amazon’s “Just Walk Out” technology enables customers at many of its Amazon Fresh and Amazon Go shops to simply pick their items and then leave whilst sensors work out which products they have taken and then automatically bill customers’ accounts.

Amazon claimed the technology was AI powered but apparently there were around 1,000 workers in India reviewing walk-out footage  to check the technology. That’s humans not machines. Amazon said that the workers were simply reviewing the system and that “this is no different than any other AI system that places a high value on accuracy, where human reviewers are common”.

If we refer back to the definition of artificial intelligence as a software’s ability to understand and solve problems but critically also learn and apply that learning, it begs the question, if 1000 workers are checking the system, is the software learning anything itself?

In the same article, the BBC’s Woollacott cites statistics from OpenOcean, an investment fund for new tech firms, revealing that while only 10% of tech start-ups mentioned using AI in their pitches in 2022, a quarter did so in 2023 and more than a third in 2024.

Back to the tech PR agency using AI to promote the new toaster. As communications advisors, we have a responsibility to ensure that what we share publicly is accurate and true and that means a careful balancing act between making our clients’ products newsworthy but being very careful not to overstate or in the worst case, completely overstep the truth.

Being an overzealous tech PR agency is one thing but working in the world of startups, we are often faced with founders whose desire for profile and funding can blind them to details like the true classification of AI. Some young tech companies believe that if they don’t say that their product is built on AI then their business isn’t innovative enough.

When faced with a passionate founder, a good tech PR agency will use journalistic rigour to analyse their client’s claims and look for the disparity between claims and capabilities. We then have to refocus the founder back on the reality of whether the product is really learning and applying to make it fit the correct definition of AI, and coach them that being apart from the crowd could even mean competitive advantage.

As a tech PR agency, how can you spot AI washing?

If you are working with new technology and need to understand if the AI claim is really true, here are a few tips on how to spot AI washing, thanks and credit to the brilliant Bernard Marr.

• Ask for details about the technology and look for terms such as neural networks, machine learning, deep learning. If a company can’t provide these specifics, it’s a red flag.
• Look for functionality that genuinely needs AI such as natural language processing.
• Ask about the company’s technology credentials like the R&D team, the developers’ skills, data scientists and related expertise.
• Question big claims like perfect results every time. AI is powerful but it’s only as good as the information it’s learning from and that’s not perfect every time.

What are the risks for tech PR agencies that get AI wrong?

According to the Consumer Protection Regulations 2008, it is an offence to provide to a consumer, false or misleading information relating to a product, the trader or any other matter relevant to a transactional decision, that is likely to deceive the average consumer. In these circumstances, a consumer has the right to claim damages in respect of any financial loss, distress or physical inconvenience or discomfort suffered.

The Competition & Markets Authority (“CMA”) can issue enforcement notices and fines of up to 10% of annual global turnover to businesses that are in breach of consumer protection law, including those practicing AI washing.

There is also the potential for investment/shareholder disputes relating to broader claims made about an organisation’s AI use and activities.

All external communications must be reviewed to ensure claims are verified, substantiated and kept under review as technologies develop.

The ASA (Advertising Standards Authority) says that a marketer should not:

• Falsely claim that a product uses AI
• Exaggerate what an AI product can do
• Claim that an AI product does something better than a non-AI product, unless this is supported by adequate evidence.

If you contract with a client that claims to be at the forefront of AI when in fact they are not, it can significantly damage your reputation with media and influencers and unwittingly mislead consumers and investors, so it is vital that you apply due diligence on the technology before signing the contract.

Finally, if in doubt, leave it out. Your duty to your client is to promote but also to protect their reputation and as such, if you have any doubts about the nature of their AI claims, it is better to leave it out of any official communication until you are confident it meets the criteria, or it has been given the green light as AI from a reputable source.