Skip to Content Skip to Content

Communications

Researchers studied how Google, OpenAI, Anthropic, and DeepSeek identify hate speech. Here’s how they vary

Researchers studied how Google, OpenAI, Anthropic, and DeepSeek identify hate speech. Here’s how they vary

A study by Yphtach Lelkes and doctoral student Neil Fasching of the Annenberg School for Communication finds that different AI models produce different definitions of hate speech for the same content, undermining consistency and predictability.

What is real about human-AI relationships?
A screenshot of a Replika chatbot screen.

Image: Courtesy of Annenberg School for Communication

What is real about human-AI relationships?

In a new paper, Annenberg School for Communication doctoral student Arelí Rocha explores how people discuss their relationships with AI chatbots.

Hailey Reissman

2 min. read

Which anti-vaping messages work for LGBTQ+ youth?

Which anti-vaping messages work for LGBTQ+ youth?

To understand how to reduce vaping among LGBTQ+ youth, researchers at Penn’s Annenberg School for Communication spoke to young Americans about anti-vaping messages on social media. They found that sexual and gender minority youth are more likely to vape, but there aren’t many anti-vaping campaigns designed specifically for them.