Ever used ChatGPT to write an email? Nothing wrong with that — unless, of course, your company’s generative AI policy prohibits it. And policies are meant to be followed, but it turns out this might not be so common when it comes to generative AI.
Salesforce recently surveyed over 14,000 workers across 14 countries to learn about unapproved AI use at work and relevant topics.
Globally, a whopping 40% of workers admit to using banned AI tools at work, while 55% admit to using “unapproved” tools. More than two-thirds of respondents also reported receiving minimal to no training on generative AI ethics, safety, or overall operations.
Those figures are similar for both Canadian and American workers, though the two countries show some deviation in generative AI behaviour and practices.
Overall, American workers reported using banned AI tools at work significantly more than Canadians (49% versus 28%). More Canadian respondents report lacking AI policies than American ones, while a small proportion of companies in both countries have “loosely defined policies.”
Generative AI offers plenty of benefits for businesses, but the speed of advancement has also sparked concern around the systems being used safely and responsibly. As a result, the Canadian government recently launched a voluntary code of conduct for how generative AI is used and developed.
So far 16 companies have signed the code, including Telus, BlackBerry, and Vector Institute. The principles agreed to include transparency in how collected information is used and human monitoring of AI systems created for public use.
Here are some more details about the report findings for Canada and the United States.
Canada findings
- 28% of Canadian workers have used banned generative AI tools at work.
- 56% of Canadian workers have used unapproved AI tools at work
Generative AI impact on productivity and engagement:
- 63% say they’re more productive with generative AI
- 45% say they’re more engaged with generative AI
Generative AI policies at work:
- 13% have loosely defined policies
- 41% have no policies
- 33% unsure
What safe AI use means to workers:
- Never using personally identifying customer data in AI tool prompts
- Never using confidential company data in AI tool prompts
- Only using company-approved generative AI tools
What ethical AI use means to workers:
- Fact-checking AI outputs before use
- Only using accuracy-validated AI tools
- Checking with relevant company departments before using AI tools
USA findings
- 49% of American workers have used banned generative AI tools at work.
- 55% of American workers have used unapproved AI tools at work
Generative AI impact on productivity and engagement:
- 72% say they’re more productive with generative AI
- 68% say they’re more engaged with generative AI
Generative AI policies at work:
- 17% have loosely defined policies
- 34% have no policies
- 22% unsure
What safe AI use means to workers:
- Never using confidential company data in AI tool prompts
- Only using company-approved generative AI tools
- Never using personally identifying customer data in AI tool prompts
What ethical AI use means to workers:
- Fact-checking AI outputs before use
- Only using accuracy-validated AI tools
- Only using AI platforms that have human oversight
Learn more about Salesforce’s AI at Work research here.
