AI washing: Misleading claims and their repercussions
Alyssa Schwartz | CPA Canada
Explaining the risks and impacts of exaggerating and misrepresenting AI capabilities in business
It’s 2024 and we’re in the midst of a modern-era gold rush. From business tools that promise advanced insights and automatic capabilities to personal wellness and entertainment—AI is booming, and investors are clamouring to get in on the action.
By the end of this decade, AI is expected to contribute $15.7-trillion USD to the global economy, according to projections from PwC. And according to Stanford’s Artificial Intelligence Index Report for 2024, the number of newly funded AI companies has surged globally to 1,812, a 40.6 per cent increase year over year.
But with great opportunity comes new risks, in this case of overstating AI’s impact or even of boasting AI-based capabilities that don’t actually exist—as in the case of Amazon’s much hyped “Just Walk Out” technology. While the company touted its checkout-free grocery service as being AI based—sensors would automatically track purchases, so shoppers could just take what they want and leave—in April, reports revealed most transactions had to be manually checked by a workforce of roughly 1000 people based in India.
According to Melissa Robertson, Principal, Research & Thought Leadership (Technology) at CPA Canada, this type of overhyping or even misrepresenting is nothing new. “This concept of washing has been on the radar a long time. We see it with things like greenwashing,” she says. “It’s the same general issue of misrepresenting what you’re doing as a business in the language you put out to customers or investors.”
Much like in the original Gold Rush, part of the problem is a general sense of lawlessness due to lack of regulations—or even standard, agreed-upon definitions of what constitutes AI. “Sometimes the technology that people are describing isn't really AI. It might be automation, it might be some other form of analytics, but it’s not AI. Where the washing comes in is promoting that you’re using it or investing in it to show you’re doing something or making progress when that’s not really the case,” Robertson says.
Laws are also slow to keep up. While the term “greenwashing” has been around since 1986, the Canadian Senate is only now looking at amendments to the Competition Act to specifically deal with misleading environmental claims. We’re nowhere near that when it comes to AI marketing. “There’s a lot of work happening (on AI regulation) … but it’s at a more basic stage,” Robertson says. “It’s more just looking at how is the technology being used and how AI systems are designed, and preventing the use of very high-risk systems without proper controls.”
But no rules doesn’t mean no consequences. Earlier this year, the SEC fined Toronto-based investment advisor Delphia (USA) Inc. and Global Predictions Inc. a total of $400,000 USD in civil penalties related to misleading statements around their use of AI.
And just like the downstream effects of greenwashing, AI washing carries other risks too. “If a company is saying they’re using (AI) and it comes to light they’re not, then obviously customers are going to question what they’re doing as an organization. It’s a reputational issue. If you’re misleading investors and they're allocating capital to these organizations because they see potential in the technology but turns out they're not actually using the technology that could be a litigation issue,” Robertson says.
“In general, organizations should be cautious with what they say they are using AI and how they are using it. As much as AI washing doesn’t have specific regulation around it, misrepresenting what you’re doing as an organization does.”