"As the Internet becomes ever more immersive for users through influencer content, generative AI, the metaverse, and beyond, we need to think more old school about child safety, perhaps counter-intuitively. What are the common-sense, practical boundaries that have long been in place in the brick and mortar world in the US to keep kids on playgrounds and playing fields, while out of strip clubs, drug markets, and other danger zones? How can we translate these trusted safeguards, which set kids up to explore in reasonably protected spaces, into the digital world kids are growing up in today and the near future?"

"A growing consensus of research findings have now concluded that large language models aren’t as capable as originally predicted and that we aren’t as close to a world-ending artificial superintelligence as doomsayers continue to insinuate. AI video generation techniques, which have seen substantial recent improvements, are probably something to watch with regard to believable disinformation and false narratives. But ultimately, it’s the simpler machine learning mechanisms, the ones that drive things like social media recommendations, that are still causing the most harm to our society."

Trust and safety work won’t go away—but the way we talk about it will: “We’re entering a quieter phase publicly, even as the work behind the scenes deepens. Political scrutiny, especially in the US, is pushing companies to retreat from public conversations about moderation and harm reduction. Some policymakers now frame inaction as virtue, claiming to protect free speech—while avoiding tough questions about digital harm. For practitioners, this means: your work still matters. But you may need to navigate tighter communications policies, shifting internal priorities, and more pressure to "stay in your lane." Don’t mistake the silence for a signal to stop—it’s a signal to get smarter about how we frame, measure, and defend what we do.”