Peter Features on CBC and AM1150 on the Perez Hilton Stream

Today, Peter featured on CBC and AM1150 to discuss the role of AI harm detection on social media platforms, after the celebrity blogger Perez Hilton livestreamed a TikTok video that appeared to show him engaging in self-harm.

While TikTok detected the event the livestream remained live for ~15 minutes, thousands of people were able to access it and share it. During Peter’s interviews he stated that AI does and can detect this sort of content, but that it isn’t necessarily robust and that in these situations seconds matter. In the AM1150 interview, Peter cautioned against fully autonomous systems that may be over or under sensitive to detecting particular forms of content and imagery, and at the same time encouraged the idea for more standards and legal requirements that describe what is expected of a platform in such situations.