Connect with us

Tech

Meta Developed 4 New Chips to Power Its AI and Recommendation Systems

Published

on

Meta Developed 4 New Chips to Power Its AI and Recommendation Systems


Meta announced Wednesday that it has developed four new computer chips that will be used to power generative AI features and content ranking systems within the tech giant’s own apps. The hardware will become part of Meta’s existing chip line known as MTIA, or Meta Training and Inference Accelerators.

Meta partnered with Broadcom to develop its latest semiconductors, which are built on top of the open-source RISC-V architecture. They’re being fabricated by Taiwan Semiconductor Manufacturing Corporation, the world’s leading chip producer.

One of the new MTIA chips, the MTIA 300, is already in production, and Meta says the other three—the MTIA 400, 450, and 500—are expected to ship sometime between early and late 2027. Putting out fresh silicon this quickly is unusual by most chip industry standards, but is essentially unheard of for a social media company that historically didn’t produce its own physical computing infrastructure.

YJ Song, a vice president of engineering at Meta, says that AI models are evolving faster than traditional chip development cycles, so AI workloads may change substantially by the time new hardware typically reaches production. “Rather than placing a bet and waiting for a long period of time, we deliberately take an iterative approach. Each MTIA generation builds on the last, using modular chiplets and incorporating the latest AI workload insights and hardware technologies,” Song said in a blog post.

The MTIA 300 will be used primarily for training algorithms that rank and recommend content to the hundreds of millions of people who use apps like Facebook and Instagram each day. The other three chips are designed to support inference, the process of running trained AI models to produce outputs like text or images.

The MTIA 400, which Meta claims delivers performance “competitive with leading commercial products,” has been tested and is expected to arrive at data centers soon. The MTIA 450 will have double the amount of high-bandwidth memory as the MTIA 400, and is supposed to ship in early 2027. Meta says the MTIA 500, which is slated to arrive later next year, will have even more memory than MTIA 450 and include “innovations in low-precision data.”

The MTIA chips are part of Meta’s broader strategy to hoard as much computing power as possible in order to develop cutting-edge artificial intelligence. Meta first shared details about its chip development plans in 2023, when it released its first product under the MTIA banner. As software companies and AI labs continue to train increasingly powerful AI models, they have begun announcing ambitious plans to build custom chips that serve their own specific AI needs. OpenAI, for example, has also said it’s partnering with Broadcom to build custom accelerators, following a path similar to Meta’s.

Earlier this year, Meta was reported to be scaling back some of its in-house efforts to make high-end chips that would compete more directly with leading players like Nvidia. The company now appears eager to dispel that narrative by announcing this new road map for MTIA chips. But making custom silicon remains enormously expensive and technically complex, which means Meta will likely continue purchasing the majority of its AI hardware from other firms, at least in the near future.

That reality is reflected in the company’s recent chip buying spree. Meta unveiled its new MTIA chips shortly after announcing multibillion dollar deals with Nvidia and AMD. The social media giant has also signed an agreement to rent chips made by Google.



Source link

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Tech

This Is the Next Wave of Political Fundraising

Published

on

This Is the Next Wave of Political Fundraising


On Monday, streamer and content creator Hasan Piker helped raise more than $56,000 in one stream for Oliver Larkin, a former Bernie Sanders campaign staffer who is seeking to primary Jared Moskowitz, a moderate Democratic congressman from Florida. It was the most the campaign had raised “in a single day,” Larkin said on X shortly after the stream ended.

Over the past few years, creators have become an essential piece of campaign messaging strategy. But Piker’s recent stream for Larkin is the latest sign that online influence is being leveraged for direct fundraising as well.

Piker isn’t alone. Trisha Paytas, a YouTuber with more than 5 million subscribers and a long history of provocative stunts, isn’t known for her political activism, but in February she donated more than $10,000 to a campaign called Creators Against ICE. The campaign, organized by the creator collective Creators for Peace, is just one in a string of fundraisers organized by coalitions of creators turning social media followings into political fundraising machines.

Unlike traditional fundraising models like super PACs that pool funds from publicly reported donors, these creator collectives pool audiences and leverage social networks and off-the-shelf tools like Shopify and Tiltify to convert followers into donors. Creators for Peace is one of the most prominent groups in a line of creator coalitions mobilizing around causes from Gaza relief to immigration aid—establishing a model that could reshape grassroots fundraising ahead of the midterm elections.

“There are a lot of creators that I think recognize the power of having a platform,” says Hassan Khadair, one of the Creators for Peace organizers. “There’s more of a call to action culturally with creators than I think there’s ever been before.”

Creators for Peace was established in 2024 by Nikki Carreon in an Instagram group DM with a handful of other creators to raise money for Gaza relief. That group chat expanded into a more than 120-person Discord server that included influencers with millions of followers on platforms like Instagram, Twitch, and YouTube. People like Kurtis Conner, Hasan Piker, and the Try Guys, who collectively boast more than 15 million followers on their primary platforms, got involved. Members shared infographics with their audiences and organized a livestream. By the end of the campaign, the group had raised more than $1.6 million.

“We largely start from zero on each new campaign. I will individually reach out to several creators, we’ll get something out, and then once we allow that to catch fire on its own, a bunch of creators will reach out to us,” says Khadair. For the Creators for Peace immigration fundraiser, Khadair says, “we really wanted to try and move out of the leftist bubble just a little bit, because a lot of our audiences tend to align with us on these issues.”

By connecting with more apolitical creators like Paytas, the Creators Against ICE campaign has raised nearly $140,000 for the National Immigration Law Center, according to the group’s Tiltify fundraiser.

Creators have come under fire for remaining silent on political issues for years. During the 2020 Black Lives Matter protests, audiences began demanding that influencers creating content on anything from fashion to food publicly speak out and take sides on political issues. In these online spaces, silence is often seen as complicity.

Groups of Democratic political influencers, like UnderTheDeskNews, have also started raising funds for whistles to alert communities about the presence of ICE agents and community watch support as well. In February, around 80 creators were part of an anti-ICE merch fundraiser tied to Bad Bunny’s Super Bowl performance, selling T-shirts, hats, and stickers featuring the singer’s Sapo Concho mascot. The campaign raised more than $100,000 for immigration legal defense funds.



Source link

Continue Reading

Tech

Iran war a melting pot for other cyber threats | Computer Weekly

Published

on

Iran war a melting pot for other cyber threats | Computer Weekly


State-backed cyber threat actors from the likes of Belarus, China and Pakistan are all ramping up their activity in the wake of the joint Israeli-US attack on Iran, even though their government paymasters are not directly involved in the war.

This is according to intelligence published by Proofpoint, which claims to have observed several such campaigns unfolding in the wild. It believes this wave of malicious activity reflects a mixture of threat actors opportunistically using the conflict to create lures in their routine options, and intelligence collection directly related to Middle Eastern governments and their allies.

“These campaigns were conducted by both known groups and previously unobserved actors, with suspected attribution to China, Belarus, Pakistan and Hamas,” wrote Proofpoint’s research team.

“The campaigns heavily relied on aspects of the conflict as topical lure content to engage the targets and often used compromised accounts belonging to government organisations to send phishing emails,” they said.

In one such campaign, Belarussian threat actor TA473, or Winter Vivern, impersonated a European Council president spokesperson relaying a statement on the European Union’s (EU’s) position on human rights, regional security and Iran’s alleged weapons of mass destruction.

It was sent to government organisations in both Europe and the Middle East – the first time Winter Vivern has been seen targeting the Middle East – and contained an HTML file which, if opened, displayed a decoy image while conducting an HTTP request in the background. However, said Proofpoint, for now at least, this request is likely intended for target tracking purposes only, as it neither observed nor retrieved any next-stage payloads.

At the same time, the China-linked UNK_InnerAmbush actor ran a phishing exercise targeting diplomats and government officials in the region. Using a compromised email address, it used the death of Ayatollah Khamenei as a lure, purporting to share “secret on-site images” obtained via the US Department of Foreign Affairs – which should be a dead giveaway to anybody with knowledge of American politics, as US foreign affairs are handled by the State Department.

Images of strikes

Days later, UNK_InnerAmbush pivoted to images of Israel’s strikes on Iran’s fossil fuel infrastructure, which have induced a major ecological disaster – but in all instances, the images were actually disguised Microsoft Shortcut (LNK) files, hosted in a password-protected ZIP or RAR archive on Google Drive. If opened, they ran executables that decrypted Cobalt Strike command and control (C2) payloads and loaded them into memory.

Meanwhile, despite their government’s non-involvement, Pakistan-aligned threat actor UNK_RobotDreams has been targeting the offices of Middle Eastern government organisations in neighbouring India, impersonating India’s Ministry of External Affairs – which is at least the correct terminology – with phishing emails purporting to advise on the security impacts of the war.

These emails contained a blurred decoy PDF attachment and a fake Adobe Reader button which, if opened, redirected to a threat actor-controlled URL that used geofencing to serve a tainted executable to its intended targets. The executable functioned as a .NET loader that retrieved a Rust backdoor from the threat actor’s C2 host via PowerShell.

“While several of these groups incorporated the war-themed lure content in operations that are largely consistent with typical targeting remits, others demonstrated a shift toward intelligence collection against Middle Eastern government and diplomatic entities,” wrote Proofpoint’s research team.

“This likely reflects an effort to gather regional intelligence on the standing, trajectory and broader geopolitical implications of the conflict. This suggests the conflict is being used both as a topical social engineering pretext and a driver of collection priorities for a range of state-aligned threat actors.”

Iran’s state APTs stirring

In contrast to the opening days of the war, during which they appeared to be lying low, leaving the virtual battlefield largely to hacktivists, Iran’s own network of state-linked threat actors is now beginning to make itself known.

Proofpoint said it had now observed TA453, or Charming Kitten, conducting phishing exercises against a US-based think tank, with its lures themed around a roundtable on air defence capabilities – although strictly speaking, this activity began before the outbreak of war.

Other Iranian threat actors, notably the Ministry of Intelligence and Security (MoIS)-linked Seedworm (aka MuddyWater, Static Kitten), have been targeting US airports, banks, non-profits and tech companies, according to intelligence from Cisco Talos.

While, as with Charming Kitten, much of this activity began in February, Cisco Talos noted the use of a previously unknown custom backdoor, dubbed Dindoor, which uses Deno – an open source JavaScript runtime – to execute.

Dindoor was first highlighted by Symantec and Carbon Black last week, and was linked to Seedworm by the use of certificates issued to aliases linked to other Seedworm malwares.

Brigid O’Gorman, senior intelligence analyst at the Symantec and Carbon Black Threat Hunter team, told our sister title, Cybersecurity Dive, that while this particular Seedworm campaign began before the current conflict, it puts the gang in a “potentially dangerous” position to be able to launch further attacks.



Source link

Continue Reading

Tech

Meta Ramps Up Efforts to Disrupt Industrialized Scamming

Published

on

Meta Ramps Up Efforts to Disrupt Industrialized Scamming


With organized, industrial-scale scamming causing a multibillion-dollar crisis around the world, Meta announced new account protections on Wednesday aimed at flagging potentially suspicious activity to users as early in a scam interaction as possible. The company also shared details about a recent Thai law enforcement collaboration that resulted in 21 arrests and Meta disabling over 150,000 user accounts associated with Southeast Asian scam compounds.

The disruptive action—a joint effort of the Royal Thai Police, the FBI, the United Kingdom’s National Crime Agency, the Australian Federal Police, and other law enforcement agencies—focused on alleged scammers targeting victims in numerous countries, including the US and UK as well as multiple Asian and Pacific region countries. The account protections Meta debuted on Wednesday include expanding its Messenger scam detection features for more users around the world, introducing warnings about potentially suspicious activity when a user is initiating a new WhatsApp device link, and testing new Facebook alerts to flag potentially suspicious friend requests.

“Transnational scam syndicates continue to exploit digital platforms and operate across multiple jurisdictions,” Gregory Kang, the deputy assistant commissioner of the Singapore Police Force, said in a statement on Wednesday. “Joint operations like this demonstrate the importance of close cooperation between law enforcement agencies and industry partners.”

Mainstream social media and communication platforms are a crucial digital meeting ground where online scammers—who are often forced laborers—and victims from around the world can cross paths. Professionalized “pig butchering”-style investment scamming has expanded in Southeast Asia and proliferated around the world, creating more urgency than ever to block and deter fraudulent activity on consumer platforms.

Meta began speaking publicly about its work focused on scam compounds at the end of 2024. That year, the company said that it had taken down more than 2 million accounts related to scam compounds.

On Wednesday, the company said that in 2025 it took down 10.9 million Facebook and Instagram accounts “associated with criminal scam centers” and removed more than 159 million scam ads across all categories. Meta has increasingly come under fire for not taking enough proactive action against scams across its platforms—with Reuters reporting in December that billions of scam ads appear everyday and internal Meta estimates forecast up to 10 percent of its revenues may come from scam advertising. A company spokesperson at the time disputed the figures. Law enforcement in many regions—including Thai and Cambodian police—have carried out a spate of operations in recent months to intervene in scam compounds, make dozens of arrests, and seize funds. And the crackdowns aren’t limited to Southeast Asia. Meta said in February, for example, that it provided support for a Nigerian Police Force and UK National Crime Agency operation focused on disrupting an alleged scam center in Nigeria.

Meta announced other efforts on Wednesday to combat scamming and abusive behavior on its platforms. The company said it is further expanding advertiser verification with a goal that 90 percent of ad revenue will come from verified advertisers by the end of 2026, which would be a major increase from 70 percent currently. The goal, Meta says, is for the final 10 percent to accommodate small, local businesses and other low-resource, benign entities that just want to run a few ads.

The company also said that its anti-scam specialists have built AI detection systems to help flag more situations where scammers may be impersonating brands, celebrities, or other public figures. These systems are also designed to catch more “deceptive links” that could be used to fool targets into visiting malicious websites.

The scamming ecosystem and industry around the world has expanded and matured to such a degree that no one platform or government can solve the problem. But experts have consistently emphasized to WIRED in recent years that Meta platforms are a key battleground where more detections and defenses could make a difference in the barrier to entry for scammers who are trying to reach new victims.

As Chris Sonderby, Meta vice president and deputy general counsel, put it in a statement on Wednesday, “we will continue to invest in technology and partnerships to stay ahead of these adversaries.”



Source link

Continue Reading

Trending