Massachusetts Senator Elizabeth Warren is the latest public figure to question how Meta is moderating content during the Israel-Hamas war. In a letter addressed to Mark Zuckerberg, Warren raises several issues reported by Instagram users since October 7, and presses Meta for more information about its underlying policies and how much content the company has taken down related to the conflict.

In the letter, Warren cites reports from the media and human rights groups about inconsistencies in the company’s moderation practices since the start of the war. In particular, she notes that numerous Instagram users have accused the company of "shadowbanning" them for posting about the conditions in Gaza. She also references a third-party audit, commissioned by Meta and published last year, that found the company violated Palestinians’ right to free expression in 2021, the last time there was a major escalation in violence in the Gaza Strip.

“Reports of Meta’s suppression of Palestinian voices raise serious questions about Meta’s content moderation practices and anti-discrimination protections,” Warren writes. “Social media users deserve to know when and why their accounts and posts are restricted, particularly on the largest platforms where vital information-sharing occurs.”

The letter asks for detailed information about how Meta is enforcing its policies in the context of the war. For example, it asks Meta to disclose statistics about the number of posts that have been removed since October 7, and how many of those takedowns have been appealed. It also asks Meta to explain reports that the company hid numerous Instagram comments with Palestinian flags for being “potentially offensive.”

The letter gives Zuckerberg a January 5 deadline to respond to the questions. Meta didn’t respond to a request for comment.

Meta has come under increasing scrutiny for its moderation practices since October 7. The company has publicly blamed several issues on unspecified technical glitches and bugs, but has also acknowledged taking temporary emergency measures to slow the spread of potentially harmful content.

Meta’s independent Oversight Board is also fast-tracking two cases related to the Israel-Hamas war content, the first time the group has opted to expedite its usual months-long process. The board said at the time it had seen a surge in appeals from Facebook and Instagram users since the start of the conflict.

Instagram inboxes are about to get a little busier. The app is adding two-second looping videos and direct replies to Notes, the status update-like feature that lives at the top of users’ inboxes.

The feature was first introduced a year ago as a way to share quick text updates with your mutual followers and close friends. Since then, the company has also added music and translation capabilities to the feature. Now, with the latest change, users will also be able to share 2-second looping videos in the space, almost like a mini Boomerang. The clip will appear at the top of your friends’ inboxes in place of your normal profile photo.

At the same time, Meta is also adding the ability for people to interact with their friends’ notes similar to how you might respond to a Stories post. Tapping on a note at the top of your inbox brings up a shortcut to reply with a GIF, sticker, voice note, photo or text. Replies to notes are routed to users’ inboxes alongside other messages.

The updates come as Meta has looked for more low-pressure ways to encourage users to share on its apps, including with smaller groups of friends. The company recently rolled out the ability to share feed posts and Reels with “close friends” only. It's also recently been spotted working on a feature called “flipside,” which could offer a finsta-like space directly in users’ profiles.

When Threads launched this summer, one of the more intriguing aspects of the new service was Meta’s promise to support ActivityPub, the open-source protocol that powers Mastodon and other decentralized social media platforms. But there were still many unanswered questions about how such interoperability would work, and just how committed Meta was to the plan.

Now, Meta is taking its first step toward making Threads compatible with the fediverse. A new test will make some Threads content available on Mastodon and other apps for the first time, Mark Zuckerberg announced in a post on Threads. “Making Threads interoperable will give people more choice over how they interact and it will help content reach more people,” he wrote. "I'm pretty optimistic about this."

It wasn’t immediately clear how this would work or how much Threads content might be available on Mastodon or other services. But the company previously introduced the ability to verify your Threads profile on Mastodon, so Meta does have some insight into Threads users who are also active in the fediverse. We’ve reached out to Meta for more details and will update if we hear back.

The test is the latest sign of growing momentum behind decentralized social media, which advocates say could upend how we interact online. Because decentralized platforms like Mastodon rely on protocols rather than a single company’s platform, more mainstream adoption of say, ActivityPub, means that people would be able to interact with a wider variety of content in more places. Meta has also said ActivityPub support would allow users to “to stop using Threads and transfer your content to another service.”

Though some fediverse enthusiasts have been skeptical about Meta’s entry into the space, Mastodon founder Eugen Rochko has said he’s supportive of their efforts. “The fact that large platforms are adopting ActivityPub is not only validation of the movement towards decentralized social media, but a path forward for people locked into these platforms to switch to better providers,” he wrote in a blog post earlier this year.

The Ray-Ban Meta smart glasses are about to get some powerful upgrades thanks to improvements to the social network’s AI assistant. The company is finally adding support for real-time information to the onboard assistant, and it’s starting to test new “multimodal” capabilities that allow it to answer questions based on your environment.

Up to now, Meta AI had a “knowledge cutoff” of December 2022, so it couldn’t answer questions about current events, or things like game scores, traffic conditions or other queries that would be especially useful while on the go. But that’s now changing, according to Meta CTO Andrew Bosworth, who said that all Meta smart glasses in the United States will now be able to access real-time info. The change is powered “in part” by Bing, he added.

Separately, Meta is starting to test one of the more intriguing capabilities of its assistant, which it’s calling “multimodal AI.” The features, first previewed during Connect, allow Meta AI to answer contextual questions about your surroundings and other queries based on what your looking at through the glasses.

Meta

The updates could go a long way toward making Meta AI feel less gimmicky and more useful, which was one of my top complaints in my initial review of the otherwise impressive smart glasses. Unfortunately, it will likely still be some time before most people with the smart glasses can access the new multimodal functionality. Bosworth said that the early access beta version will only be available in the US to a “small number of people who opt in” initially, with expanded access presumably coming sometime in 2024.

Both Mark Zuckerberg shared a few videos of the new capabilities that give an idea of what may be possible. Based on the clips, it appears users will be able to engage the feature with commands that begin with “Hey Meta, look and tell me.” Zuckerberg, for example, asks Meta AI to look at a shirt he’s holding and ask for suggestions on pants that might match. He also shared screenshots showing Meta AI identifying an image of a piece of fruit and translating the text of a meme.

In a video posted on Threads, Bosworth said that users would also be able to ask Meta AI about their immediate surroundings as well as more creative questions like writing captions for photos they just shot.

Meta’s latest update for Threads will address a long-running feature request for the company’s Twitter competitor: topic tags. The company is adding searchable tags to make it easier for people to find conversations that interest them.

Mark Zuckeberg previewed the change last month, but the feature is now available to all Threads users, according to the company. With the update, Threads users can append one tag to each post, and the app will surface tag suggestions and stats about how many other users have used the topic in the past.

Notably, though the feature can be found under the familiar # symbol, Threads’ tags are a bit different than hashtags. As Instagram head Adam Mosseri noted in a post, Threads tags can contain spaces and special characters. Threads posts are also limited to a single tag per post, so users may want to think carefully about which tag they select.

Hashtags have been a long-requested feature for Meta’s Twitter competitor, and many users have questioned why the company didn’t carry over the feature, which is also widely used on Instagram. It seems Meta has been slow to adopt the feature, however, because of fears of how it could be misused on the rapidly growing platform.

Hashtags, while long popular on Instagram and X, have also been misused. On Instagram, hashtags have been used to spread misinformation and other incredibly problematic content. And while Meta has already imposed search limitations on “potentially sensitive” content on Threads, the added restrictions on tags might discourage other kinds of abuse. “The hope is this design focuses tags more on communities and less on engagement hacking,” Mosseri explained.

But even with the limitations, the addition of tags could help make Threads more useful for tracking real-time conversations. And the fact that tags include stats about how much they’re being used suggests it could also be a precursor to some kind of trending topics feature, which has been the source of much speculation.

Meta’s Oversight Board says it will fast-track two cases dealing with content takedowns on Facebook and Instagram related to the ongoing Israel-Hamas war. The cases mark the first time the independent board has opted to expedite a review, which allows it to make a decision in as little as 48 hours instead of the typical weeks or months-long process.

The group says it has seen a surge in appeals since the start of the conflict with “an almost three-fold increase in the daily average of appeals” related to the Middle East and North Africa. The board said it selected the two cases, one from Facebook and one from Instagram, because they “address important questions relating to the conflict and represent wider issues affecting Facebook and Instagram users.”

In both cases, Meta initially removed the posts but later restored them. The case originating from Instagram stems from an early November post “showing what appears to be the aftermath of an airstrike on a yard outside Al-Shifa Hospital in Gaza City.” Meta had taken down the post, citing its rules against violent content, but restored the post with a warning screen after the Oversight Board agreed to consider the case.

The case from Facebook deals with a video of Israeli hostages filmed during the October 7 attacks in Israel. Meta removed the video, citing its dangerous organization and violence and incitement policy. According to the Oversight Board, Meta later “revised its policy guidance in response to trends in how hostage kidnapping videos were being shared and reported on,” following the October 7 attacks.

The Oversight Board said in a statement it expects to make decisions about the cases within 30 days. As with other Oversight Board cases, Meta is required to comply with the board’s decision regarding whether the appealed content should be allowed to remain on its platform. The board will also make a series of policy recommendations to the company, though Meta isn’t bound to implement those changes.

Still, the board’s recommendations in these cases will likely be watched closely as Meta has faced increased scrutiny for its content moderation decisions since the start of the conflict. The company attempted to dispel accusations that it had “shadowbanned” Instagram users for sharing posts about the conditions in Gaza. Meta later blamed some of the issues on an unspecified “bug.”

The Oversight Board has previously raised questions about the company’s handling of content related to conflicts between Israel and Hamas. Last year, an independent report, commissioned by Meta following a recommendation from the board, found discrepancies in the company’s moderation practices that violated Palestinians’ right to free expression in 2021. In response to the report, Meta said it would update several of its rules, including its Dangerous Organizations and Individuals policy.

Meta has launched a standalone version of its image generator as it tests dozens of new generative AI features across Facebook, Instagram and WhatsApp. The image generator, called Imagine, was first previewed at the company’s Connect event in November and has been available as part of Meta’s AI chatbot.

Now, with its own dedicated website at imagine.meta.com, the tool will be available outside of the company’s messaging apps. Like other generative AI tools, Imagine allows users to create images from simple text prompts. Imagine, which relies on Meta’s Emu model, will generate four images for each prompt.

The images all have a visible watermark in the lower left corner indicating they were created with Meta AI. Additionally, Meta says it will soon begin testing an invisible watermarking system that’s “resilient to common image manipulations like cropping, color change (brightness, contrast, etc.), screen shots and more.” For those interacting with the image generator in Meta’s messaging apps, the company also introduced a new “reimagine” tool, which allows users to tweak existing images created with Meta AI in chats with friends.

Interestingly, the standalone site for Imagine requires not just a Facebook or Instagram login, but a Meta account, which was introduced earlier this year so VR users could use Quest headsets without a Facebook login. It’s unclear for now if Meta planning an eventual virtual reality tie-in for Imagine, but the company has recently used its new generative AI tools try to breathe new life into its metaverse.

Meta is also testing dozens of new generative AI features across its apps. On Instagram, the company is testing the ability to convert a landscape image to portrait in Stories with a new creative tool called “Expander.” On Facebook, generative AI will also start to show up in places like Groups and Marketplace. Meta is also testing AI-generated writing suggestions for Feed posts, Facebook Dating profiles as well as AI-generated replies for creators to use in replies to Instagram direct messages.

With the latest changes, Meta is also making its 28 celebrity-infused chatbots available to all users in the United States. The company says it will test a new “long-term memory” feature for some of its AI characters so that users can more easily return to previous chats and pick up the conversation where they left off. The chatbots are available in Instagram, Messenger and WhatsApp.

The updates highlight how Meta has sought to make generative AI a core part of its service as it tries to compete with the offerings of other AI companies. Mark Zuckerberg said earlier this year that the company would bring gen AI into “every single one of our products.”

Meta is failing to stop vast networks of people using its platform to promote child abuse content, a new report in The Wall Street Journal says, citing numerous disturbing examples of child exploitation it uncovered on Facebook and Instagram. The report, which comes as Meta faces renewed pressure over its handling of children’s safety, has prompted fresh scrutiny from European Union regulators.

In the report, The Wall Street Journal detailed tests it conducted with the Canadian Centre for Child Protection showing how Meta’s recommendations can suggest Facebook Groups, Instagram hashtags and other accounts that are used to promote and share child exploitation material. According to their tests, Meta was slow to respond to reports about such content, and its own algorithms often made it easier for people to connect with abuse content and others interested in it.

For example, the Canadian Centre for Child Protection told the paper a “network of Instagram accounts with as many as 10 million followers each has continued to livestream videos of child sex abuse months after it was reported to the company." In another disturbing example, Meta initially declined to take action on a user report about a public-facing Facebook Group called “Incest.” The group was eventually taken down, along with other similar communities.

In a lengthy update on its website, Meta said that “predators are determined criminals who test app, website and platform defenses,” and that it had improved many of its internal systems to restrict “potentially suspicious adults.” The company said it had “expanded the existing list of child safety related terms, phrases and emojis for our systems to find” and had employed machine learning to uncover new search terms that could be potentially exploited by child predators.

The company said it’s using technology to identify “potentially suspicious adults” in order to prevent them from connecting with each other, including in Facebook Groups, and from seeing each other’s content in recommendations. Meta also told The Wall Street Journal it “has begun disabling individual accounts that score above a certain threshold of suspicious behavior.”

The social network is facing a growing backlash over its handling of child safety. The Wall Street Journal also recently reported that Instagram Reels recommendations are serving content aimed at people who “might have a prurient interest in children.” Dozens of states recently sued Meta for allegedly harming the mental health of its youngest users, and failing to keep children younger than 13 off its apps. Mark Zuckerberg will no doubt face intense questions about these allegations next month when he appears at a Senate Judiciary Committee hearing focused on child safety online. His counterparts from TikTok, Snap, X and Discord are also slated to testify.

Meanwhile, Meta is also facing new pressure from regulators abroad. European Union officials are using a new law to probe the company’s handling of child abuse material, following The Journal’s report. The company has been given a December 22 deadline to turn over data to the bloc.

As Meta gears up for the 2024 election, the company is grappling with a new challenge that could slow its efforts to combat foreign attempts at election interference. US government agencies have stopped sharing information with the company’s security researchers about covert influence operations on its platform.

Meta says that as of July, the government has “paused” briefings related to foreign election interference, eliminating a key source of information for the company. During a call with reporters, Meta’s head of security policy Nathaniel Gleicher, declined to speculate on the government’s motivations, but the timing lines up with a court order earlier this year that restricted the Biden Administration’s contacts with social media firms.

The order, the result of two states’ attempts to limit platforms' ability to remove misinformation, is currently suspended while the Supreme Court considers the case. But government agencies, like CISA (the Cybersecurity and Infrastructure Agency) and the FBI, have apparently opted to keep the “pause” in place.

Gleicher noted that government contacts aren’t Meta’s only source of information, and that the company continues to work with industry researchers and other civil society groups. But he acknowledged that government officials can be best-placed to advise certain kinds of threats, like those that are coordinated on other platforms. “We have seen that particularly-sophisticated threat actors, like nation states, engaged in foreign interference… there are times when government has the capability to identify these campaigns that other players may not,” he said.

Meta’s researchers regularly share details about networks of fake accounts it catches boosting foreign propaganda and conducting other kinds of influence campaigns, what the company calls “coordinated inauthentic behavior” or CIB. And while most of its takedowns don’t come as a result of government tips, the company has relied on them in detecting CIB targeting US politics. Meta acted on three separate FBI tips about fake accounts from Russia, Iran and Mexico ahead of the 2020 presidential election.

Law enforcement officials have also expressed concern about the lack of coordination with social media platforms. The FBI previously told the House Judiciary Committee that it had “discovered foreign influence campaigns on social media platforms but in some cases did not inform the companies about them because they were hamstrung by the new legal oversight,” NBC News reported, citing congressional sources.

Meta’s latest comments are the first time the company has publicly confirmed that it is no longer receiving tips about election interference. The disclosure comes as the company ramps up its efforts to prepare for multiple elections in 2024, and the inevitable attempts to manipulate political conversations on Facebook. The company said in its latest report on CIB that China is now the third-most common source of coordinated inauthentic behavior on its platform, behind Russia and Iran.

Elon Musk, facing the fact that an already financially-precarious X could be poised to lose another $75 million in ad revenue following his boosting of an antisemitic conspiracy theory, has a new message for advertisers pulling back from the platform: “Go fuck yourself.”

Musk repeated the sentiment multiple times during an appearance at The New York Times’ DealBook event. “Don’t advertise,” Musk said. “If somebody is going to try and blackmail me with advertising, blackmail me with money? Go fuck yourself. Is that clear? I hope it is.”

“Hey Bob,” Musk added, in an apparent reference to Disney CEO Bob Iger, who appeared at the same event earlier in the day and spoke about the company’s decision to pull ads following Musk’s tweet earlier this month. Iger said that Disney’s association with X was “not necessarily a positive one for us,” according to Variety.

While Musk again denied being antisemitic, he did express some regret for engaging with the tweet that’s resulted in another mass exodus of advertisers from X. “I should have not replied to that particular person… I essentially handed a loaded gun to those who hate me,” Musk said about the post, per Variety.

X didn’t immediately respond to a request for comment. X CEO Linda Yaccarino had a front-row seat to the remarks, according to The Hollywood Reporter, which reports the former ad exec sat “stone-faced” during Musk’s tirade.