Let's take a hypothetical example of a preferencing system that consistently shows content to young people that encourages them to engage in harmful behaviour and shows them information about how to harm themselves. That might be an example of a preferencing system that we would have serious concerns about because the algorithm is driving people who are still in development to consume content that is quite damaging to their health and well-being.
On May 8th, 2024. See this statement in context.