AI-Generated Indigenous Art Spurs New Ethics Rules at Canada’s Telus

(Bloomberg) -- Canadian telecommunications firm Telus Corp. is promising not to use artificial intelligence to make or copy Indigenous art, following complaints from some communities that the technology misappropriates their culture.

The announcement by Vancouver-based Telus — which sells internet, mobile phone and cable television services — underscores how businesses are seeking to maintain public trust while still exploiting the efficiencies of AI. The company’s international division also sells AI and content-moderation services to technology, media and fintech firms. 

AI-generated content that mimics Indigenous art has sparked controversies in Australia, where some artists complained that their work was being used without their permission to create pieces being sold online and others withdrew from a portrait prize competition over AI concerns, according to local news sites. In Canada, the foreign affairs ministry apologized in December for publishing an AI-generated picture to represent an Indigenous woman.

“I imagine that some of those incidents actually had spurred the conversation within the Indigenous community, and that’s probably why they brought this forward as a particular concern to us,” said Pam Snively, Telus’s chief data and trust officer. “Even before AI, I think Indigenous communities had seen imitations of their art as highly problematic and sort of an appropriation of their culture.”

Telus has trained a recommendation engine with image classification, and it uses generative AI for customer support. Staff have also used external tools like OpenAI’s Dall-E to generate images for uses like corporate slide shows, with attribution, Snively said.

Telus can’t be certain that external AI models haven’t been trained on Indigenous art, so the company’s promise is limited to its controls over the generation of images, she added.

The United Nation’s UNESCO cultural agency published a report in December with guidelines on “indigenous data sovereignty” in AI, focused on Latin America and the Caribbean. While automatically digitizing Indigenous data for “recording, transmitting and revitalizing” people’s heritage has many benefits, it also presents risks around data rights and misappropriation of aesthetic or sacred patterns, the report said.  

More stories like this are available on bloomberg.com

©2024 Bloomberg L.P.