Artificial intelligence is changing how companies snoop on each other

Open this photo in gallery:

In an apparently unprecedented power move, the intelligence chiefs from the so-called Five Eyes gathered on a stage at Stanford University in California last fall to fire a shot across the bow of China Inc. and the raging problem of intellectual property theft by state-linked firms. Citing a 1,300% increase in such investigations in the U.S. over the past several years, FBI Director Christopher Wray called the Chinese government the biggest threat to Western innovation.

But the frothy industrial espionage narrative isn’t limited to questions of national security and Chinese spies. Case in point: A little-known caveat of the Trump-era United States-Mexico-Canada Agreement (USMCA) involved far-ranging Canadian legislation criminalizing the theft of trade secrets by anyone. The move, meant to better align U.S. and Canadian law, marked what one scholar describes as a “significant overhaul of the legal landscape of trade secrets and confidential information in Canada.”

Meanwhile, new technologies and platforms have made it easier than ever to obtain sensitive corporate data subject to “trade secret” designation. The internet, not to mention the Dark Web, is awash with grey information of suspect provenance—everything from undisclosed salaries and salacious tidbits that surface in Panama Paper–type data dumps to internal corporate documents that become part of the vast universe of AI training data when someone unthinkingly uploads them to a chatbot. Certainly, the temptation by C-suites to dispatch special-ops teams to conduct oppo research on rivals has never been greater, nor more feasible.

All of which raises an intriguing governance question: How should boards, with their mandates to oversee risk management, differentiate unethical (and possibly illegal) industrial espionage from legitimate competitive intelligence (i.e., strategic foresight about markets and rival firms)?

Experts say one of the telltale signs of corporate espionage is that the information in question has been obtained unethically—a stolen thumb drive, unauthorized access to a firm’s networks, printed material lifted from the garbage outside an executive’s home and so on. Take, for example, the notorious scandal of the early 2000s involving WestJet and Air Canada, the focus of a study by Shamsud Chowdhury and Jerry Sheppard, management professors at Dalhousie’s University’s Rowe School of Business and Simon Fraser University’s Beedie School of Business, respectively.

In March 2003, a former Air Canada employee gave WestJet co-founder and top executive Mark Hill access to Air Canada’s internal network. Hill covertly checked hundreds of Air Canada “load factor” reports, detailing the number of seats sold for a given flight. WestJet, under pressure to expand its operations, used the information to undercut Air Canada. In the wake of media revelations and a lawsuit, WestJet’s stock flagged, and the company was forced to restructure its board, separating the roles of chair and CEO, both held, until then, by Clive Beddoe, one of the four founders. The two firms eventually dropped the lawsuit.

While that case is a clear instance of corporate wrongdoing—espionage seemingly endorsed by board leadership, no less—pinpointing unethical access to corporate secrets isn’t always so straightforward. As Sheppard points out, there are lots of ways companies gather information on their rivals that don’t quite fit into the realm of open-source data but isn’t theft, either. “There’s been reverse engineering of products for years,” he says, offering one example. “I don’t think that’s going to stop, and it’s not really espionage.”

Navigating this grey zone is the purview of competitive intelligence (CI) professionals. Practitioners, who generally report directly to the CEO, adhere to a strict code of ethics set out by the Strategic Consortium of Intelligence Professionals (SCIP). Jonathan Calof, a professor of international business and strategy at the University of Ottawa’s Telfer School of Management and a member of SCIP’s advisory board, is the éminence grise of Canadian CI theorists. He has a keen understanding of the many ways in which companies can work the angles and tap into networks of worldly observers, all while staying on the sunny side of SCIP’s code of ethics.

Firms involved in complicated product development processes, for instance, might think they’re keeping their cards close to the vest, but then their researchers show up at academic conferences or present papers that leave a trail of clues. “This isn’t even grey,” he says. “This is out in the open.”

Calof shares another tactic: “I can sit at a booth at a trade show and write down every question I get at my booth and who it’s coming from. I’ll see a pattern in those questions. Why is that competitor at the workshop, at the guest speaker event, asking those four questions?” A rival’s new hires can offer yet another set of clues about the company’s direction, he says.

But what’s got the CI world all excited these days is the potential use of artificial intelligence and large-language models to dig up valuable yet obscure information on competitors. A recent article in the Harvard Business Review, for example, cites the case of a northern European manufacturer that released hundreds of pages of disclosure documents, including just 14 lines on the acquisition of a parcel of land in India. While a competitor’s CI team missed that critical detail, which suggested the company’s expansion into a low-cost jurisdiction, an analysis using generative AI flagged it.

“Generative AI can become the watchful eye that spots useful insights in the field of competitive intelligence,” wrote the authors, a four-person team from universities in the U.S., the U.K and Switzerland. “And these AI-generated insights are enabling top leadership to make strategic decisions in an optimal way.”

AI on its own, however, is only half a loaf for CI professionals, warns Calof. “The big thing with AI is not collection; it’s analysis,” he says. “For it to do a good job with analysis, it’s got to have clean data. And how does it get clean data? Well, we need people who can figure out what’s good and what’s bad. If you just go to the internet and scrape it, good luck.”

The flip side is that companies should be monitoring whether their own corporate secrets have been uploaded to a chatbot, adds Sheppard: “AI makes it easier to find, and that works both ways.”

Indeed, AI is precisely where the line separating corporate espionage and competitive intelligence becomes especially blurry. Rahul Bhardwaj, CEO of the Institute of Corporate Directors, points to copyright infringement cases involving firms like The New York Times, alleging their materials have been used to train large-language models without authorization, thereby introducing unethically obtained content into generative AI searches.

While it would be difficult to argue that published news stories in a company’s CI dossier on a rival are evidence of corporate espionage, the firm’s hands may not be completely clean if it’s using AI to obtain or analyze copyrighted content. “That’s going to work its way not only through the courts,” says Bhardwaj, “but I think companies and boards will be keeping an eye on that to make sure their own competitive intelligence policies reflect that they’re going to stay on the right side of that line.”

AI notwithstanding, the evolution of the global economy—and the tech sector’s enormous strategic importance—is the development that has truly raised the stakes when it comes to snooping, ethically or otherwise, on rivals, be they companies or countries. It’s one thing for a firm to reverse engineer a dishwasher made by a competitor, notes Chowdhury, and quite another to figure out how to replicate products that might be critical links in global supply chains or infrastructure.

“It’s a very fascinating area,” he muses, “because it involves corporate governance and ethics, corporate wrongdoing, cheating, financial manipulation, and all kinds of things. Law, political science, public administration, strategy, journalism—everybody’s involved there as stakeholders.”