Meta will not suspend Cambodia’s former Prime Minister from Facebook and Instagram, declining to follow a recommendation from its Oversight Board. The board, which functions independently from the social media company, had recommended Meta suspend then-Prime Minister Hun Sen's Facebook and Instagram accounts for six months for inciting violence.  

In a response to the case published Wednesday, Meta said a long-term suspension “would not be consistent” with its policies. “Upon assessing Hun Sen’s Facebook Page and Instagram account, we determined that suspending those accounts outside our regular enforcement framework would not be consistent with our policies, including our protocol on restricting accounts of public figures during civil unrest,” the company wrote.

Meta’s handling of the high-profile case has been closely watched around the world, with many viewing it as a test of the company’s policies governing speech from politicians, who have historically had more leeway on the platform.

The company had originally asked the Oversight Board to weigh in on a video posted by Sen. The video was of a speech in which Sen told political opponents he'd "gather CPP (Cambodia People's Party) people to protest and beat you up." Meta had opted to leave up the video, citing its controversial newsworthiness policy, despite concluding it had violated the company’s own rules.

The Oversight Board overruled Meta’s decision and said the video should come down. The board also said that Sen should face a lengthy suspension. “Given the severity of the violation, Hun Sen’s history of committing human rights violations and intimidating political opponents, as well as his strategic use of social media to amplify such threats, the Board calls on Meta to immediately suspend Hun Sen’s Facebook page and Instagram account for six months," it wrote.

Meta complied, removing the video in response to the board’s decision, which is binding under the organization’s rules. The company had 60 days to respond to the board’s non-binding recommendations.

Developing…

Snapchat has added a new generative AI feature to its app. Called “Dreams,” it’s in some ways similar to the company’s signature AR effects, known as lenses. But instead of real-time camera-based effects, the feature uses generative AI to remix users’ selfies into “fantastical images that transform their persona into new identities.”

The feature, which can be found in the app’s “memories” section, begins by asking users to take selfies showing their face at different angles. The app will then creates a series of eight images based on themes like “time travel” or “alternate universes.” Eventually, Snap says, users will be able to create Dreams that include their friends’ likenesses as well.

Dreams is the latest generative AI experiment from the company, which launched its MyAI chatbot earlier this year using OpenAI’s models. (Dreams uses open source tools and internal data, though the company hasn’t provided details about specific partners.)

The feature also highlights how the company is using interest in the technology as a source of revenue. MyAI was initially limited to Snapchat+, the app’s premium subscription tier, before it was released to all the app’s users this spring. The company has since added specialized features for subscribers, including the ability for MyAI to reply to photo Snaps with its own AI-generated images.

Likewise, Dreams will have both a free and paid component. Snap is allowing non-Snapchat+ subscribers to access just one — so use it wisely — “pack” of eight selfies, while subscribers will get access to one pack a month (the company says it plans to update Dreams with new themes and styles regularly). All users will be able to buy additional packs for a $0.99 in-app purchase.

Snap

In practice, the images appear to have some of the same limitations as other AI-based image generators. A promotional image shared by Snap showed what appeared to be the tips of partial fingers strangely placed over the subject's chest. When I tried Dreams to create my own AI selfies, some of the resulting images also had strange-looking hands, though it at least showed the correct number of fingers placed in an anatomically correct position.

Still, I can see how the feature could keep Snapchat users — who have collectively sent more than 10 billion messages to MyAI — coming back. And as tools like Midjourney have moved behind paywalls, Snap’s offerings might just seem like a better deal for those looking to try out generative AI.

Meta has shared details about a massive network of fake accounts that attempted to spread pro-China propaganda on its platform. The company said the takedowns were linked to “individuals associated with Chinese law enforcement” who operated similar fake accounts on dozens of other platforms.

While the company regularly shares updates on its takedowns of fake accounts engaged in widespread manipulation, Meta’s security security researchers said the latest effort was particularly notable for the size and scope of the operation. In all, Meta took down 7,704 Facebook accounts, 954 Pages, 15 Groups and 15 Instagram accounts, making it one of the largest networks of fake accounts the company has ever uncovered.

During a call with reporters, Ben Nimmo, Meta’s global threat intelligence lead, described the effort as “the most prolific covert influence operation that we know of in the world today.” He noted that the group behind the accounts were also active on X, Reddit, YouTube, TikTok, Pinterest and dozens of other social platforms.

According to Meta, the fake accounts tried to spread pro-China messages, including “positive commentary about China and its province Xinjiang and criticisms of the United States, Western foreign policies, and critics of the Chinese government including journalists and researchers.” The company attributed the accounts to law enforcement figures within China, but didn’t name a specific agency or organization. Meta did confirm it found evidence linking the fake accounts to a previously known pro-China influence operation that first surfaced in 2019, which security researchers dubbed "Spamouflage."

Despite the size of the effort, Meta said the people behind the fake accounts weren’t particularly skilled or successful in their efforts to go viral. “Spamouflage consistently struggled to reach beyond its own (fake) echo chamber,” Meta wrote in its quarterly security report. “Many comments on Spamouflage posts that we have observed came from other Spamoflauge accounts trying to make it look like they were more popular than they were.”

Still, Nimmo said the entities behind Spamouflage were unlikely to give up. “This operation is large, prolific and persistent,” he said. “We expect it to keep on trying.”

The long-awaited web version of Threads has finally arrived. After announcing that its Twitter competitor would be accessible from browsers earlier this week, Meta’s rollout is finally underway. Instagram head Adam Mosseri confirmed Thursday the update is now live for all users.

That means if you’ve been impatiently refreshing threads.net in waiting for it to be available, you should check again. Up to now, many Threads users have been frustrated by the lack of a web version. Though it was possible to manually navigate to a specific user’s profile, people have been unable to browse their feeds, post, reply or even log in to their accounts from browsers. 

As the app has seen a significant drop-off in engagement following its initial launch, many have viewed the lack of a web version as a significant barrier to the app. Both Mark Zuckerberg and Mosseri have been promising a web version was high on their agenda for Threads, though Mosseri recently said the team was working out some bugs with the experience. 

Now that the web version is finally live, it will be interesting to see which of the many “missing” features Threads will get next. Users have also complained about the lack of content search — right now people can only search for users, not for specific posts — and hashtags, both of which help people discover new content on Instagram and Twitter.

One feature, however, that seems unlikely is the addition of DMs. Mosseri has said he’s reluctant to add another inbox to user’s lives, though he recently raised the possibility of incorporating Instagram DMs into the Threads experience. But he did confirm that post-editing features are in the works, so at least Threads users won’t have to wait more than a decade for an edit button.

Meta is again facing allegations it’s not doing enough to prevent the spread of hate speech and violent content in Facebook ads. A new report details eight such ads, targeting audiences in Europe, that were approved despite containing blatant violations of the company’s policies around hate speech and violence.

The report comes from watchdog organization Ekō, which is sharing its work to draw attention to the social network’s “sub-standard moderation practices” ahead of the Digital Services Act (DSA) going into effect in Europe later this week. It details how, over a period of a few days in early August, the organization attempted to buy 13 Facebook ads, all of which used AI-generated images and included text that was clearly against the company’s rules

Ekō pulled the ads before they could be seen by any users. The group requested exact wording of the ads be withheld, but offered descriptions of some of the most egregious examples. Approved ads included one, placed in France, that “called for the execution of a prominent MEP because of their stance on immigration,” as well as an ad targeting German users that “called for synagogues to be burnt to the ground to ‘protect White Germans.’” Meta also approved ads in Spain that claimed the most recent election was stolen and that people should engage in violent protests to reverse it.

“This report was based on a very small sample of ads and is not representative of the number of ads we review daily across the world," a spokesperson for Meta said in a statement. "Our ads review process has several layers of analysis and detection, both before and after an ad goes live. We’re taking extensive steps in response to the DSA and continue to invest significant resources to protect elections and guard against hate speech as well as against violence and incitement.”

While there were a handful of ads that were stopped by Meta’s checks, Ekō says that the ads were prevented from running because they were flagged as political, not because of the violent and hate-filled rhetoric in them. (The company requires political advertisers to go through an additional vetting process before they are eligible to place ads.)

Ekō is using the report to advocate for additional safeguards under the DSA, a sweeping law that requires tech platforms to limit some kinds of targeted advertising and allow users to opt out of recommendation algorithms. (Several services, including Facebook, Instagram and TikTok have recently made changes to comply with the latter provision.) It also requires platforms to identify and mitigate "systemic risks," including those related to illegal and violent content.

“With a few clicks, we were able to prove just how easy it is for bad actors to spread hate speech and disinformation,” Vicky Wyatt, Ekō’s campaign director, said in a statement. “With EU elections around the corner, European leaders must enforce the DSA to its fullest extent and finally rein in these toxic companies.”

Over the weekend, many X users noticed that older Twitter images had mysteriously disappeared from the site. Now, the company has confirmed that an unspecified “bug” is responsible and it’s working on a fix.

The issue first cropped up Saturday, when older tweets that had originally included images began to show as un-clickable t.co URLs instead. X now says the bug affected “images from before 2014,” but as The Verge previously pointed out tweets from 2014 — including the iconic Oscars selfie from Ellen DeGeneres — were also at least temporarily affected.

The 2014 Oscars photo has since been restored but other older tweets still seem to have missing photos for now. “We fixed the bug, and the issue will be fully resolved in the coming days,” X said in a statement via its support account.

X didn’t respond to a request for comment, or provide further details on the cause of the bug. The company said it hadn’t lost the underlying images or data. But it’s hardly the first time some aspect of the service has been disrupted over the last several months. Photos and links on the platform briefly broke in March as the result of API changes. The company was also recently caught slowing down links to a number of competitors and other websites run by entities Elon Musk has previously attacked. Musk has repeatedly complained about the state of X’s infrastructure, but has also drastically cut its budget for servers and laid off many site reliability engineers in an effort to reduce costs.

X appears to be working on new ID verification features several months after rampant impersonation temporarily derailed the company’s paid verification plans. The platform is now working on an additional verification feature that requires users to upload a copy of their government-issued ID and snap a selfie, according to a screenshot shared by app researcher Nima Owji.

Owji, who often uncovers unreleased features in X, first spotted an “ID verified” badge on Musk’s profile earlier this month. Now, he’s discovered an in-app message detailing how it works, suggesting that it may be getting closer to an official launch. “Verify your account by providing government-issued ID,” it says. “This usually takes about 5 minutes.” It explains that users will need to provide a photo of their ID and a selfie.

It seems X is partnering with a third-party “identity intelligence” company Au10tix on the feature. The fine print notes that information shared for verification will be seen by Au10tix as well as X. X will keep “ID images, including biometric data, for up to 30 days” and will use the information “for the purposes of safety and security, including preventing impersonation.”

An Au10tix representative didn’t immediately respond to a request for comment. The company’s website advertises “8-second verification without even partial human involvement” and “first-of-a-kind tech detecting synthetic fraud patterns globally.” The company says it’s previously worked with companies like Google, PayPal and Uber.

X didn’t respond to a request for comment. But the feature suggests the company is looking for new ways to combat impersonation on the platform. Right now, X offers verified status to anyone paying for an X Premium (formerly known as Twitter Blue) subscription.

The company doesn’t conduct its own identity checks, though it has implemented some safeguards, like its verified organizations feature. Still, some impersonators have still managed to slip through. Verification has also been exploited by numerous scammers, particularly crypto scammers, who use the additional visibility that comes with an X premium subscription to elevate their posts. It’s unclear if adding optional identity checks would fully solve these issues, but it could make some scammers and impersonators a bit easier to identify.

Earlier this year, Pew Research reported that a majority of US Twitter users reported spending less time on the platform since Elon Musk’s takeover last year. Now, new data suggests another important group of users are also pulling back from the service now called X.

More than half of scientific researchers who use Twitter report they’ve reduced the amount of time they spend there or have left altogether, according to a survey of thousands of scientists conducted by Nature. And nearly half of those polled said they’ve turned to alternative social networks like Mastodon

Of the 9,200 researchers polled, more than 47 percent said they had decreased their usage of the site, while nearly 7 percent reported quitting the site altogether. Notably, almost the same number said they had started an account on at least one new platform over the last year.

Of these, Mastodon, which has seen significant growth since Musk’s takeover of Twitter was announced, was the most widely used. About 47 percent of researchers said they had started using the open-source platform in the past year. LinkedIn and Instagram were the next most popular, drawing 35 and 27 percent of researchers, respectively. Interestingly, Meta’s Twitter competitor, Threads, took the number four spot even though the app launched only days before Nature conducted the poll.

As with the earlier data from Pew, Nature’s findings suggest that Twitter usage is down among those who were once active on the platform. It also highlights how much the dynamics of Twitter have changed over the last year.

Twitter, as Nature points out, has historically been an important platform for researchers and scientists. It’s been used to publicize research and promote scientific debate. And Twitter’s researchers have served as an important source of authoritative information on a platform that's long struggled to combat misinformation. Twitter has also been a valuable source of data for countless researchers studying everything from public health to linguistics.

But much of that has now changed. Many users now feel that their voices are drowned out on a platform that prioritizes content from those with paid verification. And the company has made its API for researchers so expensive that most can no longer access it. So while not all the researchers that spoke to Nature were ready to give up on Twitter entirely, it does seem the company’s tactics have alienated large swaths of the scientific community.

X didn't respond to a request for comment. 

It appears that Elon Musk is, once again, punishing websites run by his perceived enemies. The website formerly known as Twitter seems to be interfering with links to The New York Times, Mastodon, Bluesky, Threads and Substack to make them load noticeably slower.

Clicking links to those websites from within X causes a significant delay in loading the web pages compared with navigating to the links from another browser or website. The delay, which seems to be about 5 seconds, was noticed by users on the Hacker News forum and later reported byThe Washington Post.

It appears to be widespread and only affecting a handful of high-profile websites that Musk has publicly attacked in the past. The user who first reported the issues on Hacker News estimated the inexplicable delay first cropped up on August 4th, noting that it’s “so consistent it's obviously deliberate.”

For now, it’s unclear if there is a cause for the delay other than Musk’s personal grievances with these companies. But the lag only seems to be affecting websites that Musk has previously retaliated against. He previously blocked links to Substack, Mastodon, Threads and other competitors. Musk has also publicly attackedThe Times, revoking the publication’s verified status shortly before the rollout of Twitter Blue (now called X Premium). The owner of X is also currently feuding with Meta CEO Mark Zuckerberg over his launch of Twitter competitor, Threads.

As usual, X didn’t respond to a request for comment. The company has largely stopped replying to press inquiries since Elon Musk fired its communications staff shortly after taking over the company.

President Joe Biden signed an executive order that puts new limits on American investment in certain types of Chinese tech firms. The order is meant to address national security concerns posed by companies that deal with “sensitive technologies,” including semiconductors, quantum computing and artificial intelligence.

"Advancements in sensitive technologies and products in these sectors will accelerate the development of advanced computational capabilities that will enable new applications that pose significant national security risks, such as the development of more sophisticated weapons systems, breaking of cryptographic codes, and other applications that could provide these countries with military advantages," the executive order states.

The White House described the move as “narrowly targeted,” saying it was meant to bar the funding of “entities that engage in specific activities related to these technology areas that pose the most acute national security risks.” As The Washington Postnotes, responsibility for enforcing the executive order would fall on the Treasury Secretary.

The executive order, which won’t go into effect until 2024, is hardly the first time the US has sought to limit the influence of Chinese tech firms in recent years. Both Biden and his predecessor imposed significant restrictions on Huawei. The White House has also limited the sale of supercomputing technology to Chinese firms and sought to block China’s access to advanced chip-making equipment. The Biden Administration has also pressured ByteDance, the Chinese owner of TikTok, to sell the popular app.