The tech giant has positioned itself as a brand-safe alternative to other companies whose AI models have been trained on copyrighted data. But there was still a conspicuous dearth of conversation at its conference in Las Vegas this year about the risks posed by the technology, according to some attendees.