Unregulated AI usage in workplaces is a compliance nightmare

BYOAI (Bring Your Own AI) To Work is seriously common among employees, and seriously risky for businesses.

In a new collaboration from the University of Melbourne in Australia and accounting firm KPMG, a survey found that 48% of the more than 48,000 respondents from 47 countries admitted to “using AI in ways that contravene company policies,” like uploading sensitive company information to public AI tools like ChatGPT.

Last year, cybersecurity firm Cyberhaven found that the percentage of sensitive data employees fed into AI chatbots was 27%. Cyberhaven offers a product to prevent employees from leaking sensitive data to AI. Seems like the third of US employees who think AI is increasing compliance risks at work are correct.

Beyond exposing company secrets to AI behemoths like OpenAI, Google, Microsoft, and Meta, employees are also hiding their AI usage from employers. Around two in five (42%) admitted they avoid revealing when they use AI, and 45% have presented AI-generated content as their own original work.

According to two of the authors of the study—Nicole Gillespie, professor of management, and Steven Lockey, a postdoctoral research fellow, both at the Melbourne Business School—not knowing when and how employees are using AI inhibits organizations’ ability to manage the risks.

You would think companies would be cracking down on all this inappropriate use of AI at work. But only one in three employees said their workplace has a policy for using GenAI, and only 47% said they’ve received AI training, according to Gillespie and Lockey in The Conversation.

They had some sage advice and a tough conclusion for business leaders: “It takes an AI-literate workforce, robust governance and clear guidance, and a culture that supports safe, transparent and accountable use. Without these elements, AI becomes just another unmanaged liability.”

Ouch.