Meta is starting to make good on its promise to bring generative AI to all of its products. At the company’s Connect event, it revealed new AI image editing and sticker-creation features for Instagram.

A tool called "restyle" is a bit like a supercharged generative AI filter. It allows users to remix their existing photos into different looks. “Think of typing a descriptor like ‘watercolor’ or a more detailed prompt like ‘collage from magazines and newspapers, torn edges’ to describe the new look and feel of the image you want to create,” the company explained.

The company also previewed a new tool called “backdrop,” a generative AI-powered green screen feature. Users can type prompts like “put me in front of a sublime aurora borealis” will create a new version of the image with the relevant background applied. The company didn’t say when restyle and backdrop would be available to Instagram users, but said the tools would be arriving “soon.” 

Meta

It added that posts created using both backdrop and restyle would come with labels indicating "the use of AI to reduce the chances of people mistaking them for human-generated content." Meta didn't share what the labels will look like, but info screens saying "image generated by Meta AI" have previously been spotted in Instagram. 

Meta also showed off new AI-generated stickers, which will be available on Instagram, as wel as WhatsApp, Messenger and Facebook Stories. The feature allows users to quickly create custom stickers for their chats and Stories via text prompts. The stickers will be rolling out “to select English-language users” of the company’s apps “over the next month,” Meta said.

The features come as Mark Zuckerberg has repeatedly promised that generative AI features will be coming to every Meta-owned app. The company also announced more than 25 generative AI chatbots, including several personalities based on real-life celebrities and creators.

X is getting closer to releasing its long-promised alerts that will notify users about whether or not their account has been “shadowbanned.” Andrea Conway, a designer at X, offered a preview of the upcoming feature that Elon Musk first promised last year.

Conway shared two mock-ups: an alert in the notifications tab, as well as an informational page that explains why X may limit the visibility of some accounts. “We have found that your account potentially contains sensitive media — such as graphic, violent, nudity, sexual behavior, hateful symbols, or other sensitive content,” it explains.

“We may cover your posts with a warning so people who don’t want to see sensitive content can avoid it. The reach of your account and its content may also be restricted, such as being excluded from the For You and Following timelines, recommended notifications, trends, and search results.”

Underneath the message is an appeal button, so users can request X revisit its initial decision. Conway also said that users would likely be able to view their account status outside of the app’s notifications tab, but didn’t elaborate on how that might work. She added that the wording and user interface hasn't yet been finalized. 

The company previously introduced a labeling feature for individual tweets that have been “visibility limited” for violating the company’s rules. But the latest update will take that a step further as the restrictions will be visible at the account level, not just for specific tweets.

The feature touches on what has long been a hot-button issue for Twitter, and now X. The company has for years limited the reach of accounts that break its rules. Under Twitter’s previous management, the practice was known as “visibility filtering.” But the company didn’t publicly share details about the practice, or which accounts it limited, which fueled conspiracy theories about “shadowbans.”

The issue came into the spotlight again last year after Musk turned over internal emails and other company records to independent journalists, who published records of Twitter executives discussing visibility filtering. Musk later promised that a future update “will show your true account status, so you know clearly if you’ve been shadowbanned, the reason why and how to appeal.”

Of course visibility filtering and so-called “shadowbanning” were never exactly the same thing. As Recodepointed out last year, Twitter for years said that shadowbanning refers to “deliberately making someone’s content undiscoverable to everyone except the person who posted it, unbeknownst to the original poster.” Whereas visibility filtering — both under Jack Dorsey and now Musk’s leadership — hides tweets from search, recommendations and other surfaces, but doesn’t make them completely invisible.

In any case, the forthcoming update should add some additional transparency — and, potentially, more controversy — to the practice. It’s not clear when the feature will make an official debut, but Conway said the company “should have more to share on this soon.”

X, the company previously known as Twitter, could soon find itself in hot water with European Union officials due to the amount of misinformation on its platform. The platform has an outsize role in the spread of misinformation, according to a new EU report.

The EU shared its findings in its first report on platforms’ handling of mis and disinformation as part of the Digital Services Act. The sweeping law, which recently went into effect, requires major platforms to disclose details about their handling of misinformation. Dozens of companies have additionally agreed to a voluntary “Code of Practice” on disinformation. X announced in May that it was pulling out of the agreement, though the company said it would adhere to the stricter disinformation policies required under the DSA.

The report found that X outstripped many of its larger peers when it comes to the volume of disinformation on its platform, and the engagement such posts attract. “X … is the platform with the largest ratio of mis/disinformation posts,” European Commission Vice President Vera Jourova said in a statement. The report also found that X ranked the highest in discoverability of misinformation and disinformation, followed by Facebook and Instagram.

X didn’t respond to a request for comment. In a series of tweets from its Global Affairs account, the company disputed the “framing” of the data and said it remains “committed to complying with the DSA.”

In a statement, Jourova said that all of the major platforms need to “to adjust their actions to reflect that there is a war in the information space waged against us.” She said that upcoming elections within the EU “will be an important test for the Code that platforms signatories should not fail.”

She also said that Musk would not be “off the hook” just because Twitter pulled out of the code of practice, according to comments reported by The Guardian. “My message for Twitter/X is you have to comply. We will be watching what you do.”

Meta is gearing up for Connect, its annual event devoted to all things VR, AR and the metaverse. This year, the company is once again set to show off a new VR headset, with the Meta Quest 3, as well as its next-generation smart glasses.

It’s also an opportunity for Meta to showcase some of its latest AI advancements, which will reportedly come in the form of new generative AI chatbots. This year will also mark the return of an in-person component to Connect, which has been online-only since 2020. The main keynote will still be livestreamed for remote viewers, but the change will hopefully make for a more lively event than the largely pre-recorded sessions of the past few years.

Here’s a rundown of everything we’re expecting to hear about during Mark Zuckerberg’s keynote, which kicks off at 1PM ET on September 26.

Meta Quest 3 and second-gen Ray-Ban Stories update

We already know quite a bit about the Meta Quest 3, thanks to a preview from Zuckerberg earlier this year. The $499 headset has been considerably slimmed down from its predecessors and comes with redesigned controllers. It will also have better graphics and run on a new “next-gen” Qualcomm chipset, which will power new mixed-reality features akin to last year’s high-end Quest Pro. But with an official reveal at Connect, we should finally get a lot more details about the Quest 3 and its capabilities.

The VR headset likely isn’t the only Reality Labs gadget we’ll hear more about. As UploadVR noted last week, Meta CTO Andrew Bosworth recently hinted that a second-generation of Meta’s Ray-Ban-branded smart glasses are also in the works and coming "pretty soon." The new frames are expected to come with better cameras and some new features, like livestreaming, according to Janko Roettgers of the Lowpass newsletter. It’s not clear when the glasses could make an official debut — unlike the Quest 3, Meta has been fairly tight-lipped about its plans for the product — but considering Bosworth’s comments and earlier reports of a possible fall launch, it seems safe to say we’ll at least get an update on their plans.

Meta has at least one other major piece of hardware in the works, the augmented reality glasses it’s been teasing since 2020. Last year at Connect, Zuckerberg showed off a prototype for a wrist-based controller that could have AR applications, but didn’t offer a look at how the glasses themselves are progressing.

That could change this year. The Verge previously reported the company is readying a pair of smart glasses, which will use the wrist-based interface, for a 2025 release. Those glasses, however, would be something of a precursor to “full-fledged AR glasses,” which are slated for 2027. Given the timeframe, it seems more likely we’ll hear about the former, wrist-controlled frames, rather than the more ambitious AR ones, but it would make sense for the company to preview more details about the project.

Chatbots and AI

It looks like Meta is set to (once again) borrow a page from Snap’s playbook with a generative AI assistant. The Wall Street Journal reported this week that Meta is readying dozens AI chatbots with distinct personalities in a bid to attract younger users.

Zuckerberg has been teasing the idea of “AI personas” for Instagram, WhatsApp and Messenger for some time. But it now sounds as if Meta intends for these AI personalities to play a role in its VR applications as well. According to The Journal, these bots, which have names like “Alvin the Alien” and “Bob the Robot,” will also live in the company’s metaverse apps. While many aspects of the plan sound a bit cringey — and even some Meta employees are reportedly skeptical of the bots — it offers an interesting window into how Meta is using AI to boost interest in the metaverse. Which brings us to…

What about the metaverse?

The company once known as Facebook has now had nearly two years to sell the public on its vision for the metaverse. So far, that’s largely fallen flat. Reality Labs losses have ballooned to more than $20 billion since last year and are expected to climb further. Meta’s investors have been openly skeptical, while new metaverse features, like the addition of legged avatars, have proved underwhelming.

So it’s not surprising that in recent months Zuckerberg has spent more time publicly hyping the company’s AI investments than the metaverse (though he’s said the company remains committed to both). So while Meta will likely share updates on Horizon Worlds, which recently began rolling out to non-VR platforms, I expect Zuckerberg and other executives will highlight the role AI plays in its vision for the metaverse much more than they have in the past.

Though the current AI boom and rise of large-language models has drawn its share of critics, the technology has already generated much more enthusiasm than the metaverse ever has. That could end up working in Meta’s favor if it can recast some of its metaverse projects as cutting-edge AI innovation rather than a virtual world most people don’t really understand.

From alarming questions about Elon Musk’s role in Ukraine’s war effort, to new details about his complicated personal life, there has been no shortage of bombshells from Walter Isaacson’s recently released biography of Elon Musk.

The book covers his childhood in South Africa, as well as his business dealings, from his first startup to Tesla, SpaceX and Neuralink. Perhaps unsurprisingly, more than a quarter of the book is devoted to Twitter.

Isaacson spent two years with Musk and thus had a front-row seat to his takeover of Twitter, beginning with his move to become a major stakeholder last spring. While much of the drama that unfolded at Twitter (now, X) over the last year and half has been well-documented, Isaacson’s account adds telling — and at times bizarre — new details about how it all went down.

Musk almost immediately regretted his decision to buy Twitter

Isaacson describes Musk’s initial bid to buy Twitter as impulsive — the result of one of his frequent “manic” moods. And he writes that Musk regretted the plan almost immediately after the deal was put in motion — both because he thought he was overpaying, and because he was so unimpressed with Twitter’s former leadership. Musk later admits, more than once, that he bought the company because he didn’t have a choice.

“I don’t know why I did it,” he says two weeks after the deal finally closed. “The judge basically said that I have to buy Twitter or else, and now I’m like, okay, shit.”

The main motivation for increasing Twitter Blue subscriptions

We already know that Musk wants to bring banking and payments features to X, but the book makes it clear that those ambitions are very much intertwined with his push for Twitter Blue (now called X Premium) subscriptions. Isaacson writes that Musk was so focused on Twitter Blue because he saw it as a way to “get a user’s credit card information into the system, enabling Twitter someday to become the broader financial-services and payments platform.”

However, the plan was somewhat derailed by Apple, as most of Twitter’s subscribers signed up via its iPhone app, and Apple doesn’t share user data, like credit card and other financial details, with app makers. Incredibly, upon learning of this, Musk instructed Yoel Roth, Twitter’s former head of trust and safety, to “just call Apple and tell them to give you the data you need.” Roth, realizing that such a request would not go over well with Apple, declined to make the call.

Musk would later meet with Tim Cook amid a separate dispute related to Twitter’s iOS app, but, according to Isaacson, Musk opted not to bring up the user data issue. But it underscores just how important financial data is to his vision to make X an “everything app.”

Musk tried to ban the ADL and other activists in 2022

Musk often portrays himself as a noble defender of free speech, but even a sympathetic biographer is quick to point out all the ways Musk put his interests ahead of free speech after acquiring Twitter.

Months before Musk would boost the #BanTheADL hashtag, he wanted to ban the group and other activists for urging advertisers to boycott the platform, Isaacson writes. Musk apparently went to Yoel Roth, twitter’s former head of trust and safety, and demanded he “stop users from urging advertisers to boycott Twitter.”

Musk then tried to ad lib a new policy to justify what would have been an unprecedented ban. “I’m changing Twitter policy right now … blackmail is prohibited as of right now. Ban it. Ban them,” Musk said. Roth deflected and Musk apparently dropped the issue.

Musk flip-flopped on whether to restore Donald Trump’s account

Despite joking to his sons that he was buying Twitter to help Trump get reelected, Musk is no fan of Trump, according to Isaacson.

“I want to avoid the bullshit disputes about Trump,” Musk had told me a few weeks earlier, emphasizing that his principle had always been to allow free speech only if it was within the bounds of the law. “If he’s engaged in criminal activity — and it seems increasingly that he has — that’s not okay,” Musk said. “It’s not free speech to subvert democracy.”

Musk, of course, would change his mind and decide that a Twitter poll was a better way to decide the issue. Isaacson doesn’t speculate on a reason for the reversal other than saying he was in a “feisty mood” that day.

How the “hardcore Twitter” pledge came about

In the several weeks following his chaotic takeover of Twitter, Musk laid off thousands of Twitter workers. Isaacson sheds new light on how these decisions were made, writing that Musk tapped his two first cousins and their close friend — all of whom worked at Musk’s other companies — to help him identify who should be cut.

Around this time, one of the more infamous stories to come out of Twitter was an online form sent to the remaining staffers, giving them two days to commit to the new “hardcore” version of the company. According to Isaacson, the form was inspired by one of Musk’s cousins, who after digging through Twitter employees’ public Slack messages, “suggested to Musk that he give employees the chance to opt out.” Musk later decided to make the form opt-in, rather than opt-out. “We want people who declare they are hardcore.”

What really happened to Twitter’s servers

Last December, Musk decided to move thousands of Twitter servers from a data center in Sacremento to a facility in Oregon in order to save money. But when Twitter engineers said the move would take at least six months, Musk grew angry, saying he felt like the “head-explosion emoji.”

Then, two days before Christmas, Musk made an impromptu visit to the Sacramento facility and declared moving the servers didn’t “seem super hard.” By the next day, Christmas Eve, Musk had tapped his cousins and others to start the move. The group began rolling the 2,500-pound server racks, which contained “totally critical data” onto a Oregon-bound moving truck. Twitter’s rules required servers with user data to be wiped before such a move, but Musk opted to use Apple AirTags and store-bought padlocks to secure them instead.

Ultimately, all of the servers were relocated within weeks, rather than the several months the company’s engineers originally estimated. But the move also caused months of instability within Twitter’s systems. This resulted in a number of issues, including a disastrous live stream with Florida Governor Ron DeSantis, who was almost unable to announce his presidential bid.

The hasty move has also attracted the scrutiny of the FTC, which is investigating X over a number of privacy and data related issues. Recently revealed court documents refer directly to the incident as an example of the company failing to follow its own security policies.

Musk’s obsession with a mobile strategy game called Polytopia

The book is filled with details about Musk’s personal relationships. But one of the more bizarre details is his long-running obsession with a mobile strategy game called The Battle of Polytopia that many of his confidantes say is key to understanding him. Musk is apparently so addicted to the game that at one point in the narrative he skips a meeting with Tesla managers so he can keep playing.

Musk is so obsessed that he’s roped much of his inner circle, including Grimes, his brother Kimbal and Shivon Zilis, the Neuralink executive with whom he has twins, into playing as well. All of them eventually deleted the game, with his brother Kimbal saying it was “destroying” his marriage. Musk, on the other hand, deleted the app from his phone but, a couple months later, opted to keep playing.

Snapchat’s subscription service for its most dedicated users continues to grow. Snapchat+, the premium service that offers fans exclusive features, has more than 5 million paying subscribers, Snap announced Thursday.

The milestone comes after the service, which launched last June, crossed 3 million sign-ups in April. That’s still just a fraction of the 750 million people who use Snapchat each month, but it’s a significant number for the company, which has been looking to boost non-advertising sources of revenue. As Bloomberg points out, Snap CEO Evan Spiegel has said his “medium-term goal” is to reach 10 million paying users.

Snapchat has used generative AI tools and other exclusive features to lure users to signing up for the $4-a-month service. When the company first introduced its MyAI chatbot and, more recently, its generative AI selfie feature, the tools were initially limited to paying subscribers. Other perks, like the ability to check how many times friends view your Story and exclusive Bitmoji customizations, are meant to appeal to power users.

The relative success of Snapchat+ stands in contrast to X Premium (formerly known as Twitter Blue), which Elon Musk has made a centerpiece of his strategy to revitalize the embattled social media company’s business. X has not formally released subscriber counts, but one researcher, speaking to Mashable, recently estimated the number to be less than one million.

Amazon’s smart glasses are getting their first significant update in more than two years. The latest version of the Echo Frames will ship with a longer battery life and redesigned audio experience, along with several new colors and styles.

The new Echo Frames can squeeze up to six hours of media playback on one charge. They also come in several new styles for both glasses and sunglasses, including two new looks from eyewear maker Carrera (the Carrera styles will be sold as “Carrera Smart Glasses.”)

Amazon

The Alexa-enabled glasses will also come with better audio than their predecessors. Amazon says the new frames have “more balanced sound, better audio clarity, and less distortion.” The glasses also come with better onboard speech processing so the glasses can more reliably detect the wake word in varied conditions, like wind and noisy rooms.

All of the new Echo Frame will cost $269.99. The company didn’t say when they would go on sale, but people can sign up now for notifications about pre-orders.

Developing...

Meta’s paid verification service is no longer just for individuals. The social network is expanding its subscription service to businesses on Facebook, Instagram and, eventually, WhatsApp, Mark Zuckerberg announced during the company’s “Conversations” event.

Meta Verified first launched for Facebook and Instagram earlier this year, but wasn't available for those with business accounts. That’s now changing, though businesses will be paying extra for the service compared with everyone else. Verification for a single account (on either Facebook or Instagram) will cost $27.99/month for mobile sign-ups or $21.99 if the purchase is made via the web. Business owners signing up on the web can also bundle an Instagram and Facebook verification for $34.99/month. That’s significantly more than the $12- $15/month Meta Verified currently costs for individuals, but Meta is likely calculating that business owners will be more willing to shell out for the extra benefits.

Meta Verified bears some similarities to Elon Musk’s X Premium. It grants subscribers a checkmark and higher ranking in search results, as well as a fast-track to customer support channels and “proactive” protection against impersonation. The company also plans to feature "Meta Verified businesses" in its in-app recommendations on Facebook and Instagram.

Meta

When Meta Verified launches for WhatsApp, it will include a few additional benefits, like multi-device support for customer chats and the ability to create a business page that will appear in web-based search results. (Meta hasn’t said how much verified status will cost businesses on WhatsApp.)

Of course, unlike X Premium, Meta does require subscribers to verify their identities. For businesses, this means subscribers will need to provide proof of their association with the business, via a phone number or email domain. The company also requires subscribers to opt-in to two-factor authentication and meet a minimum level of activity. Meta says it will begin by making subscriptions available to business owners in “select test countries,” but declined to name the specific locations. Meta Verified for WhatsApp, on the other hand, will roll out sometime “in the future.”

In addition to the expansion of Meta Verified, the company also announced a few other updates for businesses that use its messaging tools. In-app payments on WhatsApp, which are currently available in Brazil and Singapore, will be expanding to merchants in India. The app will support credit and debit card payments, as well as other apps like Razorpay.

The company also introduced something called “WhatsApp flows” so that businesses can create customized features like appointment booking directly from in-app chats. The feature will be available to WhatsApp Business users “in the coming weeks.”

Since 2012, Yelp has caught nearly 5,000 businesses engaging in shady tactics, like paying customers for favorable ratings or hiring people to write phony reviews. Now, the company has a new tool to help people — and maybe the feds — track businesses that have tried to manipulate their standing on the review platform.

Yelp is releasing a new index that tracks every U.S establishment it’s ever caught engaging in “suspicious” activity to influence its reviews. The company has made some of this information available in the past. Yelp places temporary alerts on businesses’ pages when it discovers fake reviews, and regularly releases transparency reports detailing its moderation efforts. But the index is the first time the company has offered a single place where users can find a historical record of every business that’s ever been subject to such a warning as well as a current list of businesses with active alerts on their pages .

For Yelp, the index is both its latest move in a long-running war on fake reviews, as well as a nod to a changing regulatory environment in which fake reviews are attracting increasing scrutiny from regulators. The FTC recently proposed a formal ban on fake reviews with penalties of up to $50,000 for businesses caught buying, selling or manipulating online reviews.

Yelp has said it supports such a rule. The company’s head of user operations, Noorie Malik, points out the company has previously worked with the FTC to notify them when it discovers fake reviews and the sometimes complex operations behind them. “We'd love to get to a place where this new index develops into a regular resource for others, whether it's FTC, consumers, regulators or other sites,” Malik tells Engadget.

But she’s also quick to point out that the index is also meant to help Yelp users make “educated decisions” about where to spend their money. While you may not think much about visiting a coffee shop with a history of paying people to leave positive Yelp reviews, your feelings may be very different if you’re looking for a contractor to remodel your home, or for a daycare or moving company (all of which appear in the index).

Of course, fake reviews isn’t just a Yelp problem. Malik notes that phony reviews are often coordinated on other websites among organized groups of review rings. “We also hope that it inspires other review platforms to take a firmer stance against reduced solicitation and incentivisation,” she says.

The Supreme Court has temporarily halted a lower court ruling that prohibited White House officials from communicating with social media companies. The temporary stay, from Justice Samuel Alito, is the latest twist in a controversial attempt by two states to challenge content moderation policies at social media platforms.

The case stems from a lawsuit, brought by the attorneys general of Missouri and Louisiana, that claimed federal officials overreached in their dealings with social media companies — namely Meta, Google and Twitter — as they shaped moderation policies to handle election and COVID-19 misinformation.

A lower court had previously issued an injunction that barred White House and other federal officials from communicating with social media companies. An Appeals Court decision last week eased many of the initial restrictions, but left in place a provision that banned the surgeon general, CDC and White House officials from “pressuring” social media companies into making decisions. That order, as CNN reports, was set to take effect September 18.

That ruling is now on hold, thanks to Alito’s temporary stay, as the two sides continue to argue the case. As Bloomberg points out, the stay will be in effect until September 22, though it could be extended. 

In a filing ahead of the stay, the Solicitor General argued that “the injunction would impose grave harms on the government and the public” and that that government officials had committed no wrongdoing in their interactions with social media companies. “Rather than any pattern of coercive threats backed by sanctions, the record reflects a back-and-forth in which the government and platforms often shared goals and worked together, sometimes disagreed, and occasionally became frustrated with one another, as all parties articulated and pursued their own goals and interests during an unprecedented pandemic.”

While the current hold from Alito is another temporary measure, the case seems to be headed for a longer legal battle. The Justice Department is now laying the groundwork for a Supreme Court appeal, which could drag the case out even more.