#content-moderation

#content-moderation

Meta has been getting content moderation wrong for years, like most platforms that host user-generated content. Sometimes it's a result of deliberate design choices-privacy rollbacks, opaque policies, features that prioritize growth over safety-made even when the company knows that those choices could negatively impact users. Other times, it's simply the inevitable outcome of trying to govern billions of posts with a mix of algorithms and overstretched human reviewers.

Without directly naming SafetyCore, Google explained that the optional setting can blur photos that may contain nudity and display a warning before you view or share them. Sensitive Content Warnings appears to use SafetyCore to analyze images locally on your device. Google has emphasized that SafetyCore runs entirely on your phone -- images don't leave your device, and Google doesn't know if nudity was flagged.

Google appears to have blocked AI search results for the query "does trump show signs of dementia" as well as other questions about his mental acuity, even though it will show AI results for similar searches about other presidents. When making the search about President Trump, AI Overviews will display a message that says, "An AI Overview is not available for this search":

"I found out that I didn't get booked for a Harry Potter convention recently, as I'm now doing OnlyFans. They explained it was because it's a 'family show and OnlyFans is affiliated with porn,'" she wrote. "This was baffling to me as some actors who do conventions (most actors, actually) have done TV and films in which they've done sex scenes and nudity. I'm just playing with my hair!"

The administration's announcement of the campaign explains it will target social media platforms, short video services, the livestreaming platforms used by Chinese e-commerce sites to host infomercials and even delve into comments left across the internet. Some of the material Beijing wants to root out - such as content that incites violence - would likely fall foul of content moderators in many nations.

For decades, Mark Lemley's life as an intellectual property lawyer was orderly enough. He's a professor at Stanford University and has consulted for Amazon, Google, and Meta. "I always enjoyed that the area I practice in has largely been apolitical," Lemley tells me. What's more, his democratic values neatly aligned with those of the companies that hired him. But in January, Lemley made a radical move. "I have struggled with how to respond to Mark Zuckerberg and Facebook's descent into toxic masculinity and Neo-Nazi madness," he posted on LinkedIn. "I have fired Meta as a client."

Mastodon, an open source, decentralized alternative to X, is rolling out a somewhat controversial feature by adding quote posts, which will launch next week. The feature, which allows a user to quote someone else's post and re-share it with their own response or commentary, has contributed to a culture of " dunking " on X, where users often deride other people by responding with snark or insulting humor. To address this concern, Mastodon says it's implementing quote posts with safety controls.

Over the past 15 years, I briefed each of these companies about this complex issue. The first such briefing, back in December 2010, focused on my research on how Al-Qaeda and other terrorist groups were using YouTube to radicalize young people in the West. At Google's invitation, I met with senior Google representatives, including its head of public relations and policy, senior policy manager, senior policy counsel and free speech attorney, at the company's Washington, D.C., office.

"This is not about creator safety, customer safety, or the well-being of the industry, but rather a politicised narrative that takes advantage of a tragic personal loss," Kick told BBC News.

Instagram users are baffled after an innocuous, strawberry-adorned illustration of the words "immigrants make the country great" was flagged by the social media network and hidden by the platform's moderation tech. "We use technology or a review team to identify content that should be covered," a notice obscuring the post reads. "This post doesn't go against our Community Standards, but may contain images that some people might find upsetting." Other users encountered a "sensitive content" message on the same post. "This photo may contain graphic or violent content," it reads, even though the inoffensive post clearly doesn't.