OpenAI Starts Its Trump Strategy; CES Shows Not Everything Needs AI

Meta Platforms’ announcement last week that it had relaxed its content moderation stance immediately raised the question of whether others, such as OpenAI, would follow suit.

When I asked Chris Lehane, OpenAI’s chief global affairs officer, about that during an interview on Friday, he pointed me to a new broad-ranging policy proposal from the startup—released publicly on Monday morning. Lehane highlighted a section on users being able to personalize AI tools. 

The policy proposal says AI models should be trained to be “politically unbiased by default,” although it also says that users should have the power to “choose their own preferences to personalize their AI tools.” It adds that schools and employers have an important role in teaching citizens how to do so through classes and professional development training. 

AI is “closer to a Spotify than social media—you can pick the type of music you want to hear,” Lehane told me. “We build systems as best we can to be neutral and remove as much bias and preferences as possible, but there’s a subjective nature to that.”