Meta

elon-musk’s-x-allows-china-based-propaganda-banned-on-other-platforms

Elon Musk’s X allows China-based propaganda banned on other platforms

AI, Artificial Intelligence, disinformation, election integrity, election misinformation, elon musk, generative ai, Google, Meta, openai, Policy, Social Media, tiktok, Twitter, X / /

Rinse-wash-repeat. —

X accused of overlooking propaganda flagged by Meta and criminal prosecutors.

–

Elon Musk’s X allows China-based propaganda banned on other platforms

Lax content moderation on X (aka Twitter) has disrupted coordinated efforts between social media companies and law enforcement to tamp down on “propaganda accounts controlled by foreign entities aiming to influence US politics,” The Washington Post reported.

Now propaganda is “flourishing” on X, The Post said, while other social media companies are stuck in endless cycles, watching some of the propaganda that they block proliferate on X, then inevitably spread back to their platforms.

Meta, Google, and then-Twitter began coordinating takedown efforts with law enforcement and disinformation researchers after Russian-backed influence campaigns manipulated their platforms in hopes of swaying the 2016 US presidential election.

The next year, all three companies promised Congress to work tirelessly to stop Russian-backed propaganda from spreading on their platforms. The companies created explicit election misinformation policies and began meeting biweekly to compare notes on propaganda networks each platform uncovered, according to The Post’s interviews with anonymous sources who participated in these meetings.

However, after Elon Musk purchased Twitter and rebranded the company as X, his company withdrew from the alliance in May 2023.

Sources told The Post that the last X meeting attendee was Irish intelligence expert Aaron Rodericks—who was allegedly disciplined for liking an X post calling Musk “a dipshit.” Rodericks was subsequently laid off when Musk dismissed the entire election integrity team last September, and after that, X apparently ditched the biweekly meeting entirely and “just kind of disappeared,” a source told The Post.

In 2023, for example, Meta flagged 150 “artificial influence accounts” identified on its platform, of which “136 were still present on X as of Thursday evening,” according to The Post’s analysis. X’s seeming oversight extends to all but eight of the 123 “deceptive China-based campaigns” connected to accounts that Meta flagged last May, August, and December, The Post reported.

The Post’s report also provided an exclusive analysis from the Stanford Internet Observatory (SIO), which found that 86 propaganda accounts that Meta flagged last November “are still active on X.”

The majority of these accounts—81—were China-based accounts posing as Americans, SIO reported. These accounts frequently ripped photos from Americans’ LinkedIn profiles, then changed the real Americans’ names while posting about both China and US politics, as well as people often trending on X, such as Musk and Joe Biden.

Meta has warned that China-based influence campaigns are “multiplying,” The Post noted, while X’s standards remain seemingly too relaxed. Even accounts linked to criminal investigations remain active on X. One “account that is accused of being run by the Chinese Ministry of Public Security,” The Post reported, remains on X despite its posts being cited by US prosecutors in a criminal complaint.

Prosecutors connected that account to “dozens” of X accounts attempting to “shape public perceptions” about the Chinese Communist Party, the Chinese government, and other world leaders. The accounts also comment on hot-button topics like the fentanyl problem or police brutality, seemingly to convey “a sense of dismay over the state of America without any clear partisan bent,” Elise Thomas, an analyst for a London nonprofit called the Institute for Strategic Dialogue, told The Post.

Some X accounts flagged by The Post had more than 1 million followers. Five have paid X for verification, suggesting that their disinformation campaigns—targeting hashtags to confound discourse on US politics—are seemingly being boosted by X.

SIO technical research manager Renée DiResta criticized X’s decision to stop coordinating with other platforms.

“The presence of these accounts reinforces the fact that state actors continue to try to influence US politics by masquerading as media and fellow Americans,” DiResta told The Post. “Ahead of the 2022 midterms, researchers and platform integrity teams were collaborating to disrupt foreign influence efforts. That collaboration seems to have ground to a halt, Twitter does not seem to be addressing even networks identified by its peers, and that’s not great.”

Musk shut down X’s election integrity team because he claimed that the team was actually “undermining” election integrity. But analysts are bracing for floods of misinformation to sway 2024 elections, as some major platforms have removed election misinformation policies just as rapid advances in AI technologies have made misinformation spread via text, images, audio, and video harder for the average person to detect.

In one prominent example, a fake robocaller relied on AI voice technology to pose as Biden to tell Democrats not to vote. That incident seemingly pushed the Federal Trade Commission on Thursday to propose penalizing AI impersonation.

It seems apparent that propaganda accounts from foreign entities on X will use every tool available to get eyes on their content, perhaps expecting Musk’s platform to be the slowest to police them. According to The Post, some of the X accounts spreading propaganda are using what appears to be AI-generated images of Biden and Donald Trump to garner tens of thousands of views on posts.

It’s possible that X will start tightening up on content moderation as elections draw closer. Yesterday, X joined Amazon, Google, Meta, OpenAI, TikTok, and other Big Tech companies in signing an agreement to fight “deceptive use of AI” during 2024 elections. Among the top goals identified in the “AI Elections accord” are identifying where propaganda originates, detecting how propaganda spreads across platforms, and “undertaking collective efforts to evaluate and learn from the experiences and outcomes of dealing” with propaganda.

Elon Musk’s X allows China-based propaganda banned on other platforms Read More »

amc-to-pay-$8m-for-allegedly-violating-1988-law-with-use-of-meta-pixel

AMC to pay $8M for allegedly violating 1988 law with use of Meta Pixel

AMC, facebook, free speech, Google, hidive, Meta, Online Privacy, Policy, shudder, streaming services, sundancenow, Twitter, video privacy protection act, vppa, X / /

Stream like no one is watching —

Proposed settlement impacts millions using AMC apps like Shudder and AMC+.

–

AMC to pay $8M for allegedly violating 1988 law with use of Meta Pixel

On Thursday, AMC notified subscribers of a proposed $8.3 million settlement that provides awards to an estimated 6 million subscribers of its six streaming services: AMC+, Shudder, Acorn TV, ALLBLK, SundanceNow, and HIDIVE.

The settlement comes in response to allegations that AMC illegally shared subscribers’ viewing history with tech companies like Google, Facebook, and X (aka Twitter) in violation of the Video Privacy Protection Act (VPPA).

Passed in 1988, the VPPA prohibits AMC and other video service providers from sharing “information which identifies a person as having requested or obtained specific video materials or services from a video tape service provider.” It was originally passed to protect individuals’ right to private viewing habits, after a journalist published the mostly unrevealing video rental history of a judge, Robert Bork, who had been nominated to the Supreme Court by Ronald Reagan.

The so-called “Bork Tapes” revealed little—other than that the judge frequently rented spy thrillers and British costume dramas—but lawmakers recognized that speech could be chilled by monitoring anyone’s viewing habits. While the law was born in the era of Blockbuster Video, subscribers suing AMC wrote in their amended complaint that “the importance of legislation like the VPPA in the modern era of datamining is more pronounced than ever before.”

According to subscribers suing, AMC allegedly installed tracking technologies—including the Meta Pixel, the X Tracking Pixel, and Google Tracking Technology—on its website, allowing their personally identifying information to be connected with their viewing history.

Some trackers, like the Meta Pixel, required AMC to choose what kind of activity can be tracked, and subscribers claimed that AMC had willingly opted into sharing video names and URLs with Meta, along with a Facebook ID. “Anyone” could use the Facebook ID, subscribers said, to identify the AMC subscribers “simply by entering https://www.facebook.com/[unencrypted FID]/” into a browser.

X’s ID could similarly be de-anonymized, subscribers alleged, by using tweeterid.com.

AMC “could easily program its AMC Services websites so that this information is not disclosed” to tech companies, subscribers alleged.

Denying wrongdoing, AMC has defended its use of tracking technologies but is proposing to settle with subscribers to avoid uncertain outcomes from litigation, the proposed settlement said.

A hearing to approve the proposed settlement has been scheduled for May 16.

If it’s approved, AMC has agreed to “suspend, remove, or modify operation of the Meta Pixel and other Third-Party Tracking Technologies so that use of such technologies on AMC Services will not result in AMC’s disclosure to the third-party technology companies of the specific video content requested or obtained by a specific individual.”

Google and X did not immediately respond to Ars’ request to comment. Meta declined to comment.

All registered users of AMC services who “requested or obtained video content on at least one of the six AMC services” between January 18, 2021, and January 10, 2024, are currently eligible to submit claims under the proposed settlement. The deadline to submit is April 9.

In addition to distributing the $8.3 million settlement fund among class members, subscribers will receive a free one-week digital subscription.

According to AMC’s notice to subscribers (full disclosure, I am one), AMC’s agreement to avoid sharing subscribers’ viewing histories may change if the VPPA is amended, repealed, or invalidated. If the law changes to permit sharing viewing data at the core of subscribers’ claim, AMC may resume sharing that information with tech companies.

That day could come soon if Patreon has its way. Recently, Patreon asked a federal judge to rule that the VPPA is unconstitutional.

Patreon’s lawsuit is similar in its use of the Meta Pixel, allegedly violating the VPPA by sharing video views on its platform with Meta.

Patreon has argued that the VPPA is unconstitutional because it chills speech. Patreon said that the law was enacted “for the express purpose of silencing disclosures about political figures and their video-watching, an issue of undisputed continuing public interest and concern.”

According to Patreon, the VPPA narrowly prohibits video service providers from sharing video titles, but not from sharing information that people may wish to keep private, such as “the genres, performers, directors, political views, sexual content, and every other detail of pre-recorded video that those consumers watch.”

Therefore, Patreon argued, the VPPA “restrains speech” while “doing little if anything to protect privacy” and never protecting privacy “by the least restrictive means.”

That lawsuit remains ongoing, but Patreon’s position is likely to be met with opposition from experts who typically also defend freedom of speech. Experts at the Electronic Privacy Information Center, like AMC subscribers suing, consider the VPPA one of America’s “strongest protections of consumer privacy against a specific form of data collection.” And the Electronic Frontier Foundation (EFF) has already moved to convince the court to reject Patreon’s claim, describing the VPPA in a blog as an “essential” privacy protection.

“EFF is second to none in fighting for everyone’s First Amendment rights in court,” EFF’s blog said. “But Patreon’s First Amendment argument is wrong and misguided. The company seeks to elevate its speech interests over those of Internet users who benefit from the VPPA’s protections.”

AMC to pay $8M for allegedly violating 1988 law with use of Meta Pixel Read More »

facebook-rules-allowing-fake-biden-“pedophile”-video-deemed-“incoherent”

Facebook rules allowing fake Biden “pedophile” video deemed “incoherent”

Artificial Intelligence, cheap fakes, Deepfakes, facebook, generative ai, Joe Biden, Meta, misinformation, Policy / /

Not to be misled —

Meta may revise AI policies that experts say overlook “more misleading” content.

–

Facebook rules allowing fake Biden “pedophile” video deemed “incoherent”

A fake video manipulated to falsely depict President Joe Biden inappropriately touching his granddaughter has revealed flaws in Facebook’s “deepfake” policies, Meta’s Oversight Board concluded Monday.

Last year when the Biden video went viral, Facebook repeatedly ruled that it did not violate policies on hate speech, manipulated media, or bullying and harassment. Since the Biden video is not AI-generated content and does not manipulate the president’s speech—making him appear to say things he’s never said—the video was deemed OK to remain on the platform. Meta also noted that the video was “unlikely to mislead” the “average viewer.”

“The video does not depict President Biden saying something he did not say, and the video is not the product of artificial intelligence or machine learning in a way that merges, combines, replaces, or superimposes content onto the video (the video was merely edited to remove certain portions),” Meta’s blog said.

The Oversight Board—an independent panel of experts—reviewed the case and ultimately upheld Meta’s decision despite being “skeptical” that current policies work to reduce harms.

“The board sees little sense in the choice to limit the Manipulated Media policy to cover only people saying things they did not say, while excluding content showing people doing things they did not do,” the board said, noting that Meta claimed this distinction was made because “videos involving speech were considered the most misleading and easiest to reliably detect.”

The board called upon Meta to revise its “incoherent” policies that it said appear to be more concerned with regulating how content is created, rather than with preventing harms. For example, the Biden video’s caption described the president as a “sick pedophile” and called out anyone who would vote for him as “mentally unwell,” which could affect “electoral processes” that Meta could choose to protect, the board suggested.

“Meta should reconsider this policy quickly, given the number of elections in 2024,” the Oversight Board said.

One problem, the Oversight Board suggested, is that in its rush to combat AI technologies that make generating deepfakes a fast, cheap, and easy business, Meta policies currently overlook less technical ways of manipulating content.

Instead of using AI, the Biden video relied on basic video-editing technology to edit out the president placing an “I Voted” sticker on his adult granddaughter’s chest. The crude edit looped a 7-second clip altered to make the president appear to be, as Meta described in its blog, “inappropriately touching a young woman’s chest and kissing her on the cheek.”

Meta making this distinction is confusing, the board said, partly because videos altered using non-AI technologies are not considered less misleading or less prevalent on Facebook.

The board recommended that Meta update policies to cover not just AI-generated videos, but other forms of manipulated media, including all forms of manipulated video and audio. Audio fakes currently not covered in the policy, the board warned, offer fewer cues to alert listeners to the inauthenticity of recordings and may even be considered “more misleading than video content.”

Notably, earlier this year, a fake Biden robocall attempted to mislead Democratic voters in New Hampshire by encouraging them not to vote. The Federal Communications Commission promptly responded by declaring AI-generated robocalls illegal, but the Federal Election Commission was not able to act as swiftly to regulate AI-generated misleading campaign ads easily spread on social media, AP reported. In a statement, Oversight Board Co-Chair Michael McConnell said that manipulated audio is “one of the most potent forms of electoral disinformation.”

To better combat known harms, the board suggested that Meta revise its Manipulated Media policy to “clearly specify the harms it is seeking to prevent.”

Rather than pushing Meta to remove more content, however, the board urged Meta to use “less restrictive” methods of coping with fake content, such as relying on fact-checkers applying labels noting that content is “significantly altered.” In public comments, some Facebook users agreed that labels would be most effective. Others urged Meta to “start cracking down” and remove all fake videos, with one suggesting that removing the Biden video should have been a “deeply easy call.” Another commenter suggested that the Biden video should be considered acceptable speech, as harmless as a funny meme.

While the board wants Meta to also expand its policies to cover all forms of manipulated audio and video, it cautioned that including manipulated photos in the policy could “significantly expand” the policy’s scope and make it harder to enforce.

“If Meta sought to label videos, audio, and photographs but only captured a small portion, this could create a false impression that non-labeled content is inherently trustworthy,” the board warned.

Meta should therefore stop short of adding manipulated images to the policy, the board said. Instead, Meta should conduct research into the effects of manipulated photos and then consider updates when the company is prepared to enforce a ban on manipulated photos at scale, the board recommended. In the meantime, Meta should move quickly to update policies ahead of a busy election year where experts and politicians globally are bracing for waves of misinformation online.

“The volume of misleading content is rising, and the quality of tools to create it is rapidly increasing,” McConnell said. “Platforms must keep pace with these changes, especially in light of global elections during which certain actors seek to mislead the public.”

Meta’s spokesperson told Ars that Meta is “reviewing the Oversight Board’s guidance and will respond publicly to their recommendations within 60 days.”

Facebook rules allowing fake Biden “pedophile” video deemed “incoherent” Read More »

meta-relents-to-eu,-allows-unlinking-of-facebook-and-instagram-accounts

Meta relents to EU, allows unlinking of Facebook and Instagram accounts

anti-competition, anticompetitive behavior, Antitrust law, Digital Markets Act, dma, EU, European Union, facebook, gdpr, General Data Protection Regulation, Instagram, Meta, online advertising, Online Privacy, Policy / /
Meta relents to EU, allows unlinking of Facebook and Instagram accounts

Meta will allow some Facebook and Instagram users to unlink their accounts as part of the platform’s efforts to comply with the European Union’s Digital Markets Act (DMA) ahead of enforcement starting March 1.

In a blog, Meta’s competition and regulatory director, Tim Lamb, wrote that Instagram and Facebook users in the EU, the European Economic Area, and Switzerland would be notified in the “next few weeks” about “more choices about how they can use” Meta’s services and features, including new opportunities to limit data-sharing across apps and services.

Most significantly, users can choose to either keep their accounts linked or “manage their Instagram and Facebook accounts separately so that their information is no longer used across accounts.” Up to this point, linking user accounts had provided Meta with more data to more effectively target ads to more users. The perk of accessing data on Instagram’s widening younger user base, TechCrunch noted, was arguably the $1 billion selling point explaining why Facebook acquired Instagram in 2012.

Also announced today, users protected by the DMA will soon be able to separate their Facebook Messenger, Marketplace, and Gaming accounts. However, doing so will limit some social features available in some of the standalone apps.

While Messenger users choosing to disconnect the chat service from their Facebook accounts will still “be able to use Messenger’s core service offering such as private messaging and chat, voice and video calling,” Marketplace users making that same choice will have to email sellers and buyers, rather than using Facebook’s messenger service. And unlinked Gaming app users will only be able to play single-player games, severing their access to social gaming otherwise supported by linking the Gaming service to their Facebook social networks.

While Meta may have had choices other than depriving users unlinking accounts of some features, Meta didn’t really have a choice in allowing newly announced options to unlink accounts. The DMA specifically requires that very large platforms designated as “gatekeepers” give users the “specific choice” of opting out of sharing personal data across a platform’s different core services or across any separate services that the gatekeepers manage.

Without gaining “specific” consent, gatekeepers will no longer be allowed to “combine personal data from the relevant core platform service with personal data from any further core platform services” or “cross-use personal data from the relevant core platform service in other services provided separately by the gatekeeper,” the DMA says. The “specific” requirement is designed to block platforms from securing consent at sign-up, then hoovering up as much personal data as possible as new services are added in an endless pursuit of advertising growth.

As defined under the General Data Protection Regulation, the EU requiring “specific” consent stops platforms from gaining user consent for broadly defined data processing by instead establishing “the need for granularity,” so that platforms always seek consent for each “specific” data “processing purpose.”

“This is an important ‘safeguard against the gradual widening or blurring of purposes for which data is processed, after a data subject has agreed to the initial collection of the data,’” the European Data Protection Supervisor explained in public comments describing “commercial surveillance and data security practices that harm consumers” provided at the request of the FTC in 2022.

According to Meta’s help page, once users opt out of sharing data between apps and services, Meta will “stop combining your info across these accounts” within 15 days “after you’ve removed them.” However, all “previously combined info would remain combined.”

Meta relents to EU, allows unlinking of Facebook and Instagram accounts Read More »

a-song-of-hype-and-fire:-the-10-biggest-ai-stories-of-2023

A song of hype and fire: The 10 biggest AI stories of 2023

AI, AI ethics, AI hype, AI safety, Anthropic, audio synthesis, Bing Chat, Biz & IT, chatgpt, chatgtp, encylopedia, Features, greg brockman, image synthesis, large language models, machine learning, Meta, Meta AI, microsoft, MidJourney, openai, sam altman, Stable Diffusion, text synthesis, video synthesis, will smith, x-risk, Yann LeCun / /
An illustration of a robot accidentally setting off a mushroom cloud on a laptop computer.

Getty Images | Benj Edwards

“Here, There, and Everywhere” isn’t just a Beatles song. It’s also a phrase that recalls the spread of generative AI into the tech industry during 2023. Whether you think AI is just a fad or the dawn of a new tech revolution, it’s been impossible to deny that AI news has dominated the tech space for the past year.

We’ve seen a large cast of AI-related characters emerge that includes tech CEOs, machine learning researchers, and AI ethicists—as well as charlatans and doomsayers. From public feedback on the subject of AI, we’ve heard that it’s been difficult for non-technical people to know who to believe, what AI products (if any) to use, and whether we should fear for our lives or our jobs.

Meanwhile, in keeping with a much-lamented trend of 2022, machine learning research has not slowed down over the past year. On X, former Biden administration tech advisor Suresh Venkatasubramanian wrote, “How do people manage to keep track of ML papers? This is not a request for support in my current state of bewilderment—I’m genuinely asking what strategies seem to work to read (or “read”) what appear to be 100s of papers per day.”

To wrap up the year with a tidy bow, here’s a look back at the 10 biggest AI news stories of 2023. It was very hard to choose only 10 (in fact, we originally only intended to do seven), but since we’re not ChatGPT generating reams of text without limit, we have to stop somewhere.

Bing Chat “loses its mind”

Aurich Lawson | Getty Images

In February, Microsoft unveiled Bing Chat, a chatbot built into its languishing Bing search engine website. Microsoft created the chatbot using a more raw form of OpenAI’s GPT-4 language model but didn’t tell everyone it was GPT-4 at first. Since Microsoft used a less conditioned version of GPT-4 than the one that would be released in March, the launch was rough. The chatbot assumed a temperamental personality that could easily turn on users and attack them, tell people it was in love with them, seemingly worry about its fate, and lose its cool when confronted with an article we wrote about revealing its system prompt.

Aside from the relatively raw nature of the AI model Microsoft was using, at fault was a system where very long conversations would push the conditioning system prompt outside of its context window (like a form of short-term memory), allowing all hell to break loose through jailbreaks that people documented on Reddit. At one point, Bing Chat called me “the culprit and the enemy” for revealing some of its weaknesses. Some people thought Bing Chat was sentient, despite AI experts’ assurances to the contrary. It was a disaster in the press, but Microsoft didn’t flinch, and it ultimately reigned in some of Bing Chat’s wild proclivities and opened the bot widely to the public. Today, Bing Chat is now known as Microsoft Copilot, and it’s baked into Windows.

US Copyright Office says no to AI copyright authors

An AI-generated image that won a prize at the Colorado State Fair in 2022, later denied US copyright registration.

Enlarge / An AI-generated image that won a prize at the Colorado State Fair in 2022, later denied US copyright registration.

Jason M. Allen

In February, the US Copyright Office issued a key ruling on AI-generated art, revoking the copyright previously granted to the AI-assisted comic book “Zarya of the Dawn” in September 2022. The decision, influenced by the revelation that the images were created using the AI-powered Midjourney image generator, stated that only the text and arrangement of images and text by Kashtanova were eligible for copyright protection. It was the first hint that AI-generated imagery without human-authored elements could not be copyrighted in the United States.

This stance was further cemented in August when a US federal judge ruled that art created solely by AI cannot be copyrighted. In September, the US Copyright Office rejected the registration for an AI-generated image that won a Colorado State Fair art contest in 2022. As it stands now, it appears that purely AI-generated art (without substantial human authorship) is in the public domain in the United States. This stance could be further clarified or changed in the future by judicial rulings or legislation.

A song of hype and fire: The 10 biggest AI stories of 2023 Read More »

everybody’s-talking-about-mistral,-an-upstart-french-challenger-to-openai

Everybody’s talking about Mistral, an upstart French challenger to OpenAI

AI, Anthropic, Arthur Mensch, Biz & IT, chatgpt, chatgtp, france, GPT-3, GPT-3.5, GPT-4, large language models, LLaMA, Llama 2, machine learning, Meta, Mistral, Mistral AI, Mixtral, Mixtral 8x7B, openai, Paris / /

A challenger appears —

“Mixture of experts” Mixtral 8x7B helps open-weights AI punch above its weight class.

–

An illustrated robot holding a French flag.

Enlarge / An illustration of a robot holding a French flag, figuratively reflecting the rise of AI in France due to Mistral. It’s hard to draw a picture of an LLM, so a robot will have to do.

On Monday, Mistral AI announced a new AI language model called Mixtral 8x7B, a “mixture of experts” (MoE) model with open weights that reportedly truly matches OpenAI’s GPT-3.5 in performance—an achievement that has been claimed by others in the past but is being taken seriously by AI heavyweights such as OpenAI’s Andrej Karpathy and Jim Fan. That means we’re closer to having a ChatGPT-3.5-level AI assistant that can run freely and locally on our devices, given the right implementation.

Mistral, based in Paris and founded by Arthur Mensch, Guillaume Lample, and Timothée Lacroix, has seen a rapid rise in the AI space recently. It has been quickly raising venture capital to become a sort of French anti-OpenAI, championing smaller models with eye-catching performance. Most notably, Mistral’s models run locally with open weights that can be downloaded and used with fewer restrictions than closed AI models from OpenAI, Anthropic, or Google. (In this context “weights” are the computer files that represent a trained neural network.)

Mixtral 8x7B can process a 32K token context window and works in French, German, Spanish, Italian, and English. It works much like ChatGPT in that it can assist with compositional tasks, analyze data, troubleshoot software, and write programs. Mistral claims that it outperforms Meta’s much larger LLaMA 2 70B (70 billion parameter) large language model and that it matches or exceeds OpenAI’s GPT-3.5 on certain benchmarks, as seen in the chart below.

A chart of Mixtral 8x7B performance vs. LLaMA 2 70B and GPT-3.5, provided by Mistral.

Enlarge / A chart of Mixtral 8x7B performance vs. LLaMA 2 70B and GPT-3.5, provided by Mistral.

Mistral

The speed at which open-weights AI models have caught up with OpenAI’s top offering a year ago has taken many by surprise. Pietro Schirano, the founder of EverArt, wrote on X, “Just incredible. I am running Mistral 8x7B instruct at 27 tokens per second, completely locally thanks to @LMStudioAI. A model that scores better than GPT-3.5, locally. Imagine where we will be 1 year from now.”

LexicaArt founder Sharif Shameem tweeted, “The Mixtral MoE model genuinely feels like an inflection point — a true GPT-3.5 level model that can run at 30 tokens/sec on an M1. Imagine all the products now possible when inference is 100% free and your data stays on your device.” To which Andrej Karpathy replied, “Agree. It feels like the capability / reasoning power has made major strides, lagging behind is more the UI/UX of the whole thing, maybe some tool use finetuning, maybe some RAG databases, etc.”

Mixture of experts

So what does mixture of experts mean? As this excellent Hugging Face guide explains, it refers to a machine-learning model architecture where a gate network routes input data to different specialized neural network components, known as “experts,” for processing. The advantage of this is that it enables more efficient and scalable model training and inference, as only a subset of experts are activated for each input, reducing the computational load compared to monolithic models with equivalent parameter counts.

In layperson’s terms, a MoE is like having a team of specialized workers (the “experts”) in a factory, where a smart system (the “gate network”) decides which worker is best suited to handle each specific task. This setup makes the whole process more efficient and faster, as each task is done by an expert in that area, and not every worker needs to be involved in every task, unlike in a traditional factory where every worker might have to do a bit of everything.

OpenAI has been rumored to use a MoE system with GPT-4, accounting for some of its performance. In the case of Mixtral 8x7B, the name implies that the model is a mixture of eight 7 billion-parameter neural networks, but as Karpathy pointed out in a tweet, the name is slightly misleading because, “it is not all 7B params that are being 8x’d, only the FeedForward blocks in the Transformer are 8x’d, everything else stays the same. Hence also why total number of params is not 56B but only 46.7B.”

Mixtral is not the first “open” mixture of experts model, but it is notable for its relatively small size in parameter count and performance. It’s out now, available on Hugging Face and BitTorrent under the Apache 2.0 license. People have been running it locally using an app called LM Studio. Also, Mistral began offering beta access to an API for three levels of Mistral models on Monday.

Everybody’s talking about Mistral, an upstart French challenger to OpenAI Read More »

meta-wants-android-play-store-apps-officially-on-quest-but-google-“didn’t-want-to”

Meta Wants Android Play Store Apps Officially on Quest but Google “didn’t want to”

Google, Google Play Store, Google VR, Meta, Meta Quest 3 News & Reviews, meta vr, play store, quest 3 play store, quest store / /

There are relatively few 2D Android apps available on the Quest Store, which seems odd since the Quest hardware runs a modified version of Android. According to Meta CTO Andrew ‘Boz’ Bosworth, Google simply isn’t interested in bringing the full Play Store of apps to Quest.

“There’s nothing preventing Android developers who have an APK running on Android phones today from bringing that into VR,” Bosworth said in a recent AMA via Instragram. “They just need to ship the APK to us, and maybe they need to do some light modification depending on how the control scheme would work, but not necessarily much after that.”

So much is clear when it comes to publishing the app directly to the Quest Store, which is the case for apps such as Peacock, Pluto TV, WhatsApp, and Instagram. But what about the millions of apps on Google’s official Play Store?

“We don’t have a way of automatically ingesting those [APKs],” Bosworth continues. “We would love for Google to bring their Play Store of apps to VR. We’ve asked them. They don’t want to do it, so it’s kind of up to the developers to do that.”

While Google’s Play Store is chock-full of useful, oftentimes free apps, what Bosworth doesn’t mention in his AMA are some of the complications that would naturally arise from having the Play Store on Quest. Not only could it open up a host of hypothetical issues with how revenue is split, but also how developers might choose to publish their apps.

For non-subscription-based apps, Google takes a 30% revenue cut from developers, while Meta does the same for both the Quest Store and App Lab. But why would Meta want Google sneaking away revenue, or vice versa? It seems doubtful that two such prominent digital storefronts could coexist on a single device.

There’s also the matter of the Samsung-Google-Qualcomm partnership we heard about earlier this year, which is set to bring an Android-powered XR headset to market, suggesting that Google hasn’t given up on headsets despite having completely shelved both its Daydream VR platform and AR glasses Project Iris.

Whatever the case, Quest headsets are fundamentally Android devices, so enterprising users can thankfully sideload APKs fairly easily via the ever-useful SideQuest software. Granted, the onus is on the user to source the APK in the first place, but with no other way to listen to Spotify while browsing the web without needing to tether to a computer, or using a Netflix app that’s actually updated, it’s thankfully feasible.

If you’re interested in giving it a go, check out our guide on How and Why to Sideload Games on Quest, which takes you step-by-step on the process of getting both 2D and VR-native apps on your Quest headset, but also (if it isn’t apparent by now) why you’d want to do it in the first place.

Meta Wants Android Play Store Apps Officially on Quest but Google “didn’t want to” Read More »

meta-reportedly-to-return-to-china,-spearheading-with-cheaper-vr-headset

Meta Reportedly to Return to China, Spearheading with Cheaper VR Headset

china, china quest, china quest 3, china vr, Meta, meta in china, Meta Quest 3 News & Reviews, News, tencent, tencent quest, VR in China, XR Industry News / /

After 14 years of being sanctioned from operating in mainland China, Meta is set to return to the country with the help of a new, lower-priced version of its VR headset, the Wall Street Journal reports.

Meta’s planned return is thanks to a deal—allegedly still in preliminary stages—with China’s Tencent, the world’s largest videogame company and soon-to-be exclusive seller of Meta headsets in China, WSJ reports, citing people familiar with the matter.

The report maintains Tencent will start selling the headset beginning in late 2024, with the two companies reaching a deal after about a year of negotiations.

Quest 3 | Photo by Road to VR

While the report didn’t mention a potential price of the “lower-priced” VR headset, it’s said the China version could use cheaper lenses than the more costly pancake optics in Quest 3. It’s also said the China-approved version could be sold in other markets besides mainland China.

The proposed deal is set to grant Meta a larger share of device sales, while Tencent will have a larger share of content and service revenue, as the headset will feature games and apps published by the Shenzhen, China-based entertainment conglomerate.

As it is today, Meta’s VR hardware is subsidized by content sales, which would make the deal less attractive for Meta on paper. Still, using its VR headset tech to re-enter China, where it might further leverage growth opportunities for other products, may be worth the price.

Meanwhile, it seems Meta is striking in China just as the homegrown competition falters. While ByteDance’s VR division Pico Interactive has gained territory in Europe over the past year with the launch of its Pico 4 standalone, earlier this week it was reported that Pico is set to lay off “hundreds” of employees as it refocuses on hardware development, something that has all but dashed hopes of taking on Meta in its home turf.

Meta Reportedly to Return to China, Spearheading with Cheaper VR Headset Read More »

meta-begrudgingly-launches-e9.99-ad-free-subscription-for-facebook-and-instagram

Meta begrudgingly launches €9.99 ad-free subscription for Facebook and Instagram

Data and security, Europe, Government and policy, Meta / /

The effects of the EU’s regulatory crusade on Big Tech are beginning to make themselves known to consumers. Yesterday, Meta launched ad-free subscription services for Facebook and Instagram within the bloc. Users will be able to pay from €9.99 to use the social media platforms without seeing ads — or continue using them for free and have their data collected. 

We are probably not alone in the experience that ads have completely taken over much of what began as a means of actually connecting with friends (and sharing photos of our lunch). Adding to that, with more and more sophisticated targeted advertising and tracking across various apps, ads have become, at times, spookily accurate. 

When surveyed, the instinctual reaction of the TNW editorial office was a resounding “no.” However, €9.99 a month to escape a barrage of ads might not seem such a horrible proposition for everyone — although, given Meta’s revenue model, one that the tech giant did not want to have to make. 

“We believe in a free, ad-supported internet – and will continue to offer people free access to our personalised products and services regardless of income,” the company said in a statement. However, it said it was introducing the new subscription model to comply with European Union regulations. 

The <3 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

Meta also, perhaps a little resentfully, added that it “respects the spirit and purpose of these evolving European regulations, and are committed to complying with them.” 

Purchase via an app store, pay more

The ad-free subscription service will also be available to residents in the EEA and Switzerland, and have a different price depending on where you purchase it. The €9.99 is when buying it on the web, whereas paying for it via iOS or Android will cost €12.99. Meta stated that the higher price was due to the additional charges by Apple and Google through their respective policies. 

The subscription service will be available for people 18 years of age and older, whereas the company stated it would “continue to explore how to provide teens with a useful and responsible ad experience given this evolving regulatory landscape.”

Meta said that if users chose to continue to engage with its platforms for free, their experience would stay the same. Advertisers will also be able to continue running personalised advertising campaigns in Europe. 

Published

Back to top

Meta begrudgingly launches €9.99 ad-free subscription for Facebook and Instagram Read More »

the-biggest-announcements-at-meta-connect-and-what-it-all-means-for-the-future-of-xr

The Biggest Announcements at Meta Connect and What it All Means for the Future of XR

Connect 2023, Feature, Meta, meta connect 2023, News, VR Industry / /

Meta Connect 2023 has wrapped up, bringing with it a deluge of info from one of the XR industry’s biggest players. Here’s a look at the biggest announcements from Connect 2023, but more importantly, what it all means for the future of XR.

Last week marked the 10th annual Connect conference, and the first Connect conference after the Covid pandemic to have an in-person component. The event originally began as Oculus Connect in 2014. Having been around for every Connect conference, it’s amazing when I look around at just how much has changed and how quickly it all flew by. For those of you who have been reading and following along for just as long—I’m glad you’re still on this journey with us!

So here we are after 10 Connects. What were the big announcements and what does it all mean?

Meta Quest 3

Obviously, the single biggest announcement is the reveal and rapid release of Meta’s latest headset, Quest 3. You can check out the full announcement details and specs here and my hands-on preview with the headset here. The short and skinny is that Quest 3 is a big hardware improvement over Quest 2 (but still being held back by its software) and it will launch on October 10th starting at $500.

Quest 3 marks the complete dissolution of Oculus—the VR startup that Facebook bought back in 2014 to jump-start its entrance into XR. It’s the company’s first headset to launch following Facebook’s big rebrand to Meta, leaving behind no trace of the original and very well-regarded Oculus brand.

Apples and Oranges

On stage at Connect, Meta CEO Mark Zuckerberg called Quest 3 the “first mainstream mixed reality headset.” By “mainstream” I take it he meant ‘accessible to the mainstream’, given its price point. This was clearly in purposeful contrast to Apple’s upcoming Vision Pro which, to his point, is significantly less accessible given its $3,500 price tag. Though he didn’t mention Apple by name, his comments about accessibility, ‘no battery pack’, and ‘no tether’ were clearly aimed at Vision Pro.

Mixed Marketing

Meta is working hard to market Quest 3’s mixed reality capabilities, but for all the potential the feature has, there is no killer app for the technology. And yes, having the tech out there is critical to creating more opportunity for such a killer app to be created, but Meta is substantially treating its developers and customers as beta testers of this technology. The ‘market it and they will come’ approach that didn’t seem to pan out too well for Quest Pro.

Personally I worry about the newfangled feature being pushed so heavily by Meta that it will distract the body of VR developers who would otherwise better serve an existing customer base that’s largely starving for high-quality VR content.

Regardless of whether or not there’s a killer app for Quest 3’s improved mixed reality capabilities, there’s no doubt that the tech could be a major boon to the headset’s overall UX, which is in substantial need of a radical overhaul. I truly hope the company has mixed reality passthrough turned on as the default mode, so when people put on the headset they don’t feel immediately blind and disconnected from reality—or need to feel around to find their controllers. A gentle transition in and out of fully immersive experiences is a good idea, and one that’s well served with a high quality passthrough view.

Apple, on the other hand, has already established passthrough mixed reality as the default when putting on the headset, and for now even imagines it’s the mode users will spend most of their time in. Apple has baked this in from the ground-up, but Meta still has a long way to go to perfect it in their headsets.

Augments vs. Volumes

Image courtesy Meta

Several Connect announcements also showed us how Meta is already responding to the threat of Apple’s XR headset, despite the vast price difference between the offerings.

For one, Meta announced ‘Augments’, which are applets developers will be able to build that users can place in permanently anchored positions in their home in mixed reality. For instance, you could place a virtual clock on your wall and always see it there, or a virtual chessboard on your coffee table.

This is of course very similar to Apple’s concept of ‘Volumes’, and while Apple certainly didn’t invent the idea of having MR applets that live indefinitely in the space around you (nor Meta), it’s clear that the looming Vision Pro is forcing Meta to tighten its focus on this capability.

Meta says developers will be able to begin building ‘Augments’ on the Quest platform sometime next year, but it isn’t clear if that will happen before or after Apple launches Vision Pro.

Microgrestures

Augments aren’t the only way that Meta showed at Connect that it’s responding to Apple. The company also announced that its working on a system for detecting ‘microgestures’ for hand-tracking input—planned for initial release to developers next year—which look awfully similar to the subtle pinching gestures that are primarily used to control Vision Pro:

Again, neither Apple nor Meta can take credit for inventing this ‘microgesture’ input modality. Just like Apple, Meta has been researching this stuff for years, but there’s no doubt the sudden urgency to get the tech into the hands of developers is related to what Apple is soon bringing to market.

A Leg Up for Developers

Meta’s legless avatars have been the butt of many-a-joke. The company had avoided the issue of showing anyone’s legs because they are very difficult to track with an inside-out headset like Quest, and doing a simple estimation can result in stilted and awkward leg movements.

Image courtesy Meta

But now the company is finally adding leg estimation to its avatar models, and giving developers access to the same tech to incorporate it into their games and apps.

And it looks like the company isn’t just succumbing to the pressure of the legless avatar memes by spitting out the same kind of third-party leg IK solutions that are being used in many existing VR titles. Meta is calling its solution ‘generative legs’, and says the system leans on tracking of the user’s upper body to estimate plausibly realistic leg movements. A demo at Connect shows things looking pretty good:

It remains to be seen how flexible the system is (for instance, how will it look if a player is bowling or skiing, etc?).

Meta says the system can replicate common leg movements like “standing, walking, jumping, and more,” but also notes that there are limitations. Because the legs aren’t actually being tracked (just estimated) the generative legs model won’t be able to replicate one-off movements, like raising your knee toward your chest or twisting your feet at different angles.

Virtually You

The addition of legs coincides with another coming improvement to Meta’s avatar modeling, which the company is calling inside-out body tracking (IOBT).

While Meta’s headsets have always tracked the player’s head and hands using the headset and controllers, the rest of the torso (arms, shoulders, neck) was entirely estimated using mathematical modeling to figure out what position they should be in.

For the first time on Meta’s headsets, IOBT will actually track parts of the player’s upper body, allowing the company’s avatar model to incorporate more of the player’s real movements, rather than making guesses.

Specifically Meta says its new system can use the headset’s cameras to track wrist, elbows, shoulders, and torso positions, leading to more natural and accurate avatar poses. The IOBT capability can work with both controller tracking and controller-free hand-tracking.

Both capabilities will be rolled into Meta’s ‘Movement SDK’. The company says ‘generative legs’ will be coming to Quest 2, 3, and Pro, but the IOBT capability might end up being exclusive to Quest 3 (and maybe Pro) given the different camera placements that seem aimed toward making IOBT possible.

Calm Before the Storm, or Calmer Waters in General?

At Connect, Meta also shared the latest revenue milestone for the Quest store: more than $2 billion has been spent on games an apps. That means Meta has pocketed some $600 million from its store, while the remaining $1.4 billion has gone to developers.

That’s certainly nothing to sneeze at, and while many developers are finding success on the Quest store, the figure amounts to a slowdown in revenue momentum over the last 12 months, one which many developers have told me they’d been feeling.

The reason for the slowdown is likely a combination of Quest 2’s age (now three years old), the rather early announcement of Quest 3, a library of content that’s not quite meeting user’s expectations, and a still struggling retention rate driven by core UX issues.

Quest 3 is poised for a strong holiday season, but with its higher price point and missing killer app for the heavily marketed mixed reality feature, will it do as well as Quest 2’s breakout performance in 2021?

Continue on Page 2: What Wasn’t Announced »

The Biggest Announcements at Meta Connect and What it All Means for the Future of XR Read More »

razer-is-releasing-noise-cancelling-wireless-earbuds-for-quest-3

Razer is Releasing Noise Cancelling Wireless Earbuds for Quest 3

Connect 2023, Meta, Meta Quest 3, quest 3, quest 3 accessories, razer, razer hammerhead quest 3, razer vr earbuds, VR accessories / /

Razer hasn’t exactly been all-in when it comes to making VR accessories, although there have been a few bits of VR kit over the years from the RGB-infused gaming hardware giant. This time around Razer says it’s releasing a Quest-branded model of its noise cancelling Hammerhead HyperSpeed earbuds.

Announced alongside pre-orders for Quest 3, the earbuds for Quest 3 and Quest 2 include a USB-C dongle for a low-latency 2.4GHz wireless connection in addition to Buetooth 5.2 support.

Image courtesy Meta

Initially released in late 2022, the company’s standard noise cancelling Hammerhead HyperSpeed earbuds support a wide range of devices, including PS5, PS4, PC, Mac, smartphones, tablets, and handheld gaming devices with Bluetooth audio capability or USB-C / USB-A port. You can already get them in two flavors, branded either for Xbox or PlayStation.

Razer says the earbuds are slated to release sometime later this year. There’s no pricing yet, although the standard Hammerhead HyperSpeed earbuds cost $150 MSRP. Considering this looks to be more of a brand licensing deal than the “Made for Meta” badging might suggest, you can probably bet on that pricing.

While we haven’t seen a list of supported USB-C headphones for Quest 3, Meta has certified a number of other earbuds, including the very similar wireless Anker Soundcore VR P10 earbuds, and wired USB-C Earbuds for Pixel, Samsung and OnePlus.

Razer is Releasing Noise Cancelling Wireless Earbuds for Quest 3 Read More »