Social Networks

TikTok's Traffic Bounces Back Despite Being Pulled Off App Stores (cnbc.com) 17

Despite being removed from app stores and facing a potential U.S. ban, TikTok has regained nearly 90% of its user traffic, according to Cloudflare Radar. "DNS traffic for TikTok-related domains has continued to recover since service restoration, and is currently about 10% lower than pre-shutdown level," said David Belson, head of data insight at Cloudflare. CNBC reports: The data from Cloudflare shows that, for the most part, TikTok has managed to maintain the bulk of its users and creators in the U.S. despite going offline for about 14 hours and remaining off of the Apple or Google app stores.

As for its alternatives, Cloudflare's data shows a spike in traffic the day of the temporary ban, with levels remaining steadily higher in the following week. Traffic for alternatives began to grow a week ahead of the expected shutdown, driven by the increased popularity of RedNote, known as Xiaohongshu in China, Belson said.

But traffic to TikTok alternatives peaked on Jan. 19, the day TikTok returned online, he added. "DNS traffic fell rapidly once the shutdown ended, and has continued to slowly decline over the last week and a half," Belson said.

AI

Taiwan Says Government Departments Should Not Use DeepSeek, Citing Security Concerns (reuters.com) 37

An anonymous reader shares a report: Taiwan's digital ministry said on Friday that government departments should not use Chinese startup DeepSeek's artificial intelligence (AI) service, saying that as the product is from China it represents a security concern.

Democratically-governed Taiwan has long been wary of Chinese tech given Beijing's sovereignty claims over the island and its military and political threats against the government in Taipei. In a statement, Taiwan's Ministry of Digital Affairs said that government departments are not allowed to use DeepSeek's AI service to "prevent information security risks".

"DeepSeek's AI service is a Chinese product, and its operation involves cross-border transmission and information leakage and other information security concerns, and is a product that jeopardises the country's information security," the ministry said.

Privacy

Italy Blocks DeepSeek Over Data Privacy Concerns (reuters.com) 30

Italy's data protection agency has blocked the Chinese AI chatbot DeekSeek after its developers failed to disclose how it collects user data or whether it is stored on Chinese servers. Reuters reports: DeepSeek could not be accessed on Wednesday in Apple or Google app stores in Italy, the day after the authority, known also as the Garante, requested information on its use of personal data. In particular, it wanted to know what personal data is collected, from which sources, for what purposes, on what legal basis and whether it is stored in China. The authority's decision -- aimed at protecting Italian users' data -- came after the Chinese companies that supply chatbot service to DeepSeek provided information that "was considered to totally insufficient," the authority said in a note on its website. The Garante added that the decision had "immediate effect" and that it had also opened an investigation. Thanks to new submitter axettone for sharing the news.
AI

Has Europe's Great Hope For AI Missed Its Moment? (ft.com) 39

France's Mistral AI is facing mounting pressure over its future as an independent European AI champion, as competition intensifies from U.S. tech giants and China's emerging players. The Paris-based startup, valued at $6.5 billion and backed by Microsoft and Nvidia, has struggled to keep pace with larger rivals despite delivering advanced AI models with a fraction of their resources.

The pressure increased this week after China's DeepSeek released a cutting-edge open-source model that challenged Mistral's efficiency-focused strategy. Mistral CEO Arthur Mensch dismissed speculation about selling to Big Tech companies, saying the firm hopes to go public eventually. However, one investor told the Financial Times that "they need to sell themselves."

The stakes are high for Europe's tech ambitions. Mistral remains the region's only significant player in large language models, the technology behind ChatGPT, after Germany's Aleph Alpha pivoted away from the field last year. The company has won customers including France's defense ministry and BNP Paribas, but controls just 5% of the enterprise AI market compared to OpenAI's dominant share.
AI

India Lauds Chinese AI Lab DeepSeek, Plans To Host Its Models on Local Servers (techcrunch.com) 11

India's IT minister on Thursday praised DeepSeek's progress and said the country will host the Chinese AI lab's large language models on domestic servers, in a rare opening for Chinese technology in India. From a report: "You have seen what DeepSeek has done -- $5.5 million and a very very powerful model," IT Minister Ashwini Vaishnaw said on Thursday, responding to criticism New Delhi has received for its own investment in AI, which has been much less than many other countries.

Since 2020, India has banned more than 300 apps and services linked to China, including TikTok and WeChat, citing national security concerns. The approval to allow DeepSeek to be hosted in India appears contingent on the platform storing and processing all Indian users' data domestically, in line with India's strict data localization requirements. [...] DeepSeek's models will likely be hosted on India's new AI Compute Facility. The facility is powered by 18,693 graphics processing units (GPUs), nearly double its initial target -- almost 13,000 of those are Nvidia H100 GPUs, and about 1,500 are Nvidia H200 GPUs.

AI

After DeepSeek Shock, Alibaba Unveils Rival AI Model That Uses Less Computing Power (venturebeat.com) 59

Alibaba has unveiled a new version of its AI model, called Qwen2.5-Max, claiming benchmark scores that surpass both DeepSeek's recently released R1 model and industry standards like GPT-4o and Claude-3.5-Sonnet. The model achieves these results using a mixture-of-experts architecture that requires significantly less computational power than traditional approaches.

The release comes amid growing concerns about China's AI capabilities, following DeepSeek's R1 model launch last week that sent Nvidia's stock tumbling 17%. Qwen2.5-Max scored 89.4% on the Arena-Hard benchmark and demonstrated strong performance in code generation and mathematical reasoning tasks. Unlike U.S. companies that rely heavily on massive GPU clusters -- OpenAI reportedly uses over 32,000 high-end GPUs for its latest models -- Alibaba's approach focuses on architectural efficiency. The company claims this allows comparable AI performance while reducing infrastructure costs by 40-60% compared to traditional deployments.
Security

Chinese and Iranian Hackers Are Using US AI Products To Bolster Cyberattacks (msn.com) 19

Hackers linked to China, Iran and other foreign governments are using new AI technology to bolster their cyberattacks against U.S. and global targets, according to U.S. officials and new security research. WSJ: In the past year, dozens of hacking groups in more than 20 countries turned to Google's Gemini chatbot to assist with malicious code writing, hunts for publicly known cyber vulnerabilities and research into organizations to target for attack, among other tasks, Google's cyber-threat experts said. While Western officials and security experts have warned for years about the potential malicious uses of AI, the findings released Wednesday from Google are some of the first to shed light on how exactly foreign adversaries are leveraging generative AI to boost their hacking prowess.

This week, the China-built AI platform DeepSeek upended international assumptions about how far along Beijing might be the AI arms race, creating global uncertainty about a technology that could revolutionize work, diplomacy and warfare. Expand article logo Continue reading Groups with known ties to China, Iran, Russia and North Korea all used Gemini to support hacking activity, the Google report said. They appeared to treat the platform more as a research assistant than a strategic asset, relying on it for tasks intended to boost productivity rather than to develop fearsome new hacking techniques. All four countries have generally denied U.S. hacking allegations.

AI

White House 'Looking Into' National Security Implications of DeepSeek's AI 53

During the first press briefing of Donald Trump's second administration, White House press secretary, Karoline Leavitt, said that the National Security Council was "looking into" the potential security implications of China's DeepSeek AI startup. Axios reports: DeepSeek's low-cost but highly advanced models have shaken the consensus that the U.S. had a strong lead in the AI race with China. Responding to a question from Axios' Mike Allen, Leavitt said President Trump saw this as a "wake-up call" for the U.S. AI industry, but remained confident "we'll restore American dominance." Leavitt said she had personally discussed the matter with the NSC earlier on Tuesday.

In the combative tone that characterized much of her first briefing, Leavitt claimed the Biden administration "sat on its hands and allowed China to rapidly develop this AI program," while Trump had moved quickly to appoint an AI czar and loosen regulations on the AI industry.
Leavitt also commented on the mysterious drones spotted flying around New Jersey at the end of last year, saying they were "authorized to be flown by the FAA."
Earth

Atomic Scientists Adjust 'Doomsday Clock' Closer Than Ever To Midnight (reuters.com) 162

The Bulletin of Atomic Scientists moved their Doomsday Clock to 89 seconds before midnight on Tuesday, the closest to catastrophe in the timepiece's 78-year history. The Chicago-based group cited Russia's nuclear threats during its Ukraine invasion, growing tensions in the Middle East, China's military pressure near Taiwan, and the rapid advancement of AI as key factors. The symbolic clock, created in 1947 by scientists including Albert Einstein, moved one second closer than last year's setting.
AI

Nvidia Dismisses China AI Threat, Says DeepSeek Still Needs Its Chips 77

Nvidia has responded to the market panic over Chinese AI group DeepSeek, arguing that the startup's breakthrough still requires "significant numbers of NVIDIA GPUs" for its operation. The US chipmaker, which saw more than $600 billion wiped from its market value on Monday, characterized DeepSeek's advancement as "excellent" but asserted that the technology remains dependent on its hardware.

"DeepSeek's work illustrates how new models can be created using [test time scaling], leveraging widely-available models and compute that is fully export control compliant," Nvidia said in a statement Monday. However, it stressed that "inference requires significant numbers of NVIDIA GPUs and high-performance networking." The statement came after DeepSeek's release of an AI model that reportedly achieves performance comparable to those from US tech giants while using fewer chips, sparking the biggest one-day drop in Nvidia's history and sending shockwaves through global tech stocks.

Nvidia sought to frame DeepSeek's breakthrough within existing technical frameworks, citing it as "a perfect example of Test Time Scaling" and noting that traditional scaling approaches in AI development - pre-training and post-training - "continue" alongside this new method. The company's attempt to calm market fears follows warnings from analysts about potential threats to US dominance in AI technology. Goldman Sachs earlier warned of possible "spillover effects" from any setbacks in the tech sector to the broader market. The shares stabilized somewhat in afternoon trading but remained on track for their worst session since March 2020, when pandemic fears roiled markets.
Facebook

Meta Sets Up War Rooms To Analyze DeepSeek's Tech (businessinsider.com) 39

Meta has set up four war rooms to analyze DeepSeek's technology, including two focusing on how High-Flyer reduced training costs, and one on what data High-Flyer may have used, The Information's Kalley Huang and Stephanie Palazzolo report. China's DeepSeek is a large-language open source model that claims to rival offerings from OpenAI's ChatGPT and Meta Platforms, while using a much smaller budgets.
AI

A New Bid for TikTok from Perplexity AI Would Give the US Government a 50% Stake (apnews.com) 113

An anonymous reader shared this report from the Associated Press: Perplexity AI has presented a new proposal to TikTok's parent company that would allow the U.S. government to own up to 50% of a new entity that merges Perplexity with TikTok's U.S. business, according to a person familiar with the matter... The new proposal would allow the U.S. government to own up to half of that new structure once it makes an initial public offering of at least $300 billion, said the person, who was not authorized to speak about the proposal. The person said Perplexity's proposal was revised based off of feedback from the Trump administration. If the plan is successful, the shares owned by the government would not have voting power, the person said. The government also would not get a seat on the new company's board.

Under the plan, ByteDance would not have to completely cut ties with TikTok, a favorable outcome for its investors. But it would have to allow a "full U.S. board control," the person said.

Under the proposal, the China-based tech company would contribute TikTok's U.S. business without the proprietary algorithm that fuels what users see on the app, according to a document seen by the Associated Press.

United States

New CIA Director Touts 'Low Confidence' Assessment About Covid Lab Leak Theory (cnn.com) 196

Slashdot reader DevNull127 writes: "Every US intelligence agency still unanimously maintains that Covid-19 was not developed as a biological weapon," CNN reported today.

But what about the possibility of an accidental leak (rather than Covid-19 originating in wild animal meat from the Wuhan Market)? "The agency has for years said it did not have enough information to determine which origin theory was more likely."

CNN notes there's suddenly been a new announcement "just days" after the CIA's new director took the reins — former lawyer turned Republican House Representative John Ratcliffe. While the market-origin theory remains a possibility according to the CIA, CNN notes that Ratcliffe himself "has long favored the theory that the pandemic originated from research being done in China and vowed in an interview published in Breitbart on Thursday that he would make the issue a Day 1 priority."

"We have low confidence in this judgement," the CIA says in the complete text of its announcement, "and will continue to evaluate any available credible new intelligence reporting or open-source information that could change CIA's assessment."

After speaking to a U.S. official, CNN added these details about the assessment: It was not made based on new intelligence gathered by the US government — officials have long said such intelligence is unlikely to surface so many years later — and instead was reached after a review of existing information.

"CIA continues to assess that both research-related and natural origin scenarios of the COVID-19 pandemic remain plausible," a CIA spokesperson said in a statement Saturday.

CNN adds that "Many scientists believe the virus occurred naturally in animals and spread to humans in an outbreak at a market in Wuhan, China...."
Social Networks

Oracle and US Investors (Including Microsoft) Discuss Taking Control of TikTok in the US (npr.org) 53

A plan to keep TikTok available in the U.S. "involves tapping software company Oracle and a group of outside investors," reports NPR, "to effectively take control of the app's global operations, according to two people with direct knowledge of the talks..."

"[P]otential investors who are engaged in the talks include Microsoft." Under the deal now being negotiated by the White House, TikTok's China-based owner ByteDance would retain a minority stake in the company, but the app's algorithm, data collection and software updates will be overseen by Oracle, which already provides the foundation of TikTok's web infrastructure... "The goal is for Oracle to effectively monitor and provide oversight with what is going on with TikTok," said the person directly involved in the talks, who was not authorized to speak publicly about the deliberations. "ByteDance wouldn't completely go away, but it would minimize Chinese ownership...." Officials from Oracle and the White House held a meeting on Friday about a potential deal, and another meeting has been scheduled for next week, according to the source involved in the discussions, who said Oracle is interested in a TikTok stake "in the tens of billions," but the rest of the deal is in flux...

Under a law passed by Congress and upheld by the Supreme Court, TikTok must execute what is known as "qualified divestiture" from ByteDance in order to stay in business in the U.S... A congressional staffer involved in talks about TikTok's future, who was not authorized to speak publicly, said binding legal agreements from the White House ensuring ByteDance cannot covertly manipulate the app will prove critical in winning lawmakers' approval. "A key part is showing there is no operational relationship with ByteDance, that they do not have control," the Congressional staffer said. "There needs to be no backdoors where China can potentially gain access...."

Chinese regulators, who have for years opposed the selling of TikTok, recently signaled that they would not stand in the way of a TikTok ownership change, saying acquisitions "should be independently decided by the enterprises and based on market principles." The statement, at first, does not seem to say much, but negotiators in the White House believe it indicates that Beijing is not planning to block a deal that gives American investors a majority-stake position in the company.

"Meanwhile, Apple and Google still have not returned TikTok to app stores..."
United States

America Lags on Renewable Energy. Blame Regulations and Grid Connection Issues (msn.com) 127

"For years, renewable energy proponents have hoped to build a U.S. electric grid powered by wind, solar, geothermal and — to a lesser extent — nuclear power..." writes the Washington Post. In America's power markets "the economics of clean energy are strong," with renewable energy cheaper than fossil fuel plants in many jurisdictions.

But the Post spoke to the "electricity modeling" director at nonpartisan clean energy think tank Energy Innovation, who offered this assessment. "The technology is ready, and the financial services are ready — but the question nobody really put enough thought into was, could the government keep up? And at the moment, the answer is no." [R]enewable developers say that the new technologies are stymied by complicated local and federal regulations, a long wait to connect to the electricity grid, and community opposition... "The U.S. offshore wind business is at a very nascent stage versus Europe or China," Rob Barnett, a senior analyst at Bloomberg Intelligence, said in an email. "With the new permitting pause, it's doubtful much progress for this emerging industry will be made...." After the Inflation Reduction Act passed, Rhodium Group — an independent clean energy research firm — estimated that between 2023 and 2025, on average, the country would add between 36 and 46 gigawatts of clean electricity to the grid every year. Late last year, however, the group found that the country only installed around 27 gigawatts in 2023. The U.S.'s renewable growth is now expected to fall on the low end of that range — or miss it entirely.

"It actually is really hard to build a lot of this stuff fast," said Trevor Houser, partner in climate and energy at Rhodium Group. As a result, Rhodium found, the country only cut carbon emissions by 0.2 percent in 2024... A significant amount of this lag has come from wind power, where problems with supply chains and getting permits and approval to build has put a damper on development. But solar construction is also on the low end of what experts were expecting...

Developers point to lags in the interconnection queue — a system that gives new solar, wind or fossil fuel projects permission to connect to the larger electricity grid. According to a report from Lawrence Berkeley National Laboratory, it can now take nearly 3 years for a project to get through the queue. The grid operator that covers the Mid-Atlantic and parts of the Midwest, PJM, had over 3,300 projects in its queue at the end of 2023. The vast majority of these applications are for renewables — more than the entire number of active wind farms in the nation... There are possible solutions. Some developers hope to reuse old fossil fuel sites, like coal plants, that are already connected to the grid — bypassing the long queue entirely. The Federal Energy Regulatory Commission has instated new rules to make it easier to build transmission lines.

Part of the problem is that wind and solar facilities "sometimes need to be built hundreds or even thousands of miles away" — requiring long transmission lines. Sandhya Ganapathy, CEO of EDP Renewables North America, tells the Post that in America, "The grid that we have was never designed to handle this kind of load." And yet last year just 255 miles of new transmission line were built in the U.S., according to the American Clean Power Association. And Ganapathy also complains that approval for a new renewable energy project takes "anywhere between six to eight years" — which makes developers hesitant to build. "Why are we taking a big risk of a massive investment if I will not be able to sell the electrons?"

The end result? The Washington Post writes that "Experts once hoped that by the end of the decade the United States could generate up to 80 percent of its power with clean power... Now, some wonder if the country will be able to reach even 60 percent."
United States

Scale AI CEO Says China Has Quickly Caught the US With DeepSeek 79

The U.S. may have led China in the AI race for the past decade, according to Alexandr Wang, CEO of Scale AI, but on Christmas Day, everything changed. From a report: Wang, whose company provides training data to key AI players including OpenAI, Google and Meta , said Thursday at the World Economic Forum in Davos, Switzerland, that DeepSeek, the leading Chinese AI lab, released an "earth-shattering model" on Christmas Day, then followed it up with a powerful reasoning-focused AI model, DeepSeek-R1, which competes with OpenAI's recently released o1 model.

"What we've found is that DeepSeek ... is the top performing, or roughly on par with the best American models," Wang said. In an interview with CNBC, Wang described the artificial intelligence race between the U.S. and China as an "AI war," adding that he believes China has significantly more Nvidia H100 GPUs -- AI chips that are widely used to build leading powerful AI models -- than people may think, especially considering U.S. export controls. [...] "The United States is going to need a huge amount of computational capacity, a huge amount of infrastructure," Wang said, later adding, "We need to unleash U.S. energy to enable this AI boom."
DeepSeek's holding company is a quant firm, which happened to have a lot of GPUs for trading and mining. DeepSeek is their "side project."
Power

Chinese Fusion Reactor Maintains Steady State For Almost 18 Minutes (charmingscience.com) 50

Longtime Slashdot readers smooth wombat and AmiMoJo shares a fusion energy breakthrough from China. Charming Science reports: China's "artificial sun," officially known as the Experimental Advanced Superconducting Tokamak (EAST), has achieved a groundbreaking milestone in fusion energy research. According to the Chinese Academy of Sciences (CAS), EAST recently sustained high-confinement plasma operation for an unprecedented 1,066 seconds, shattering the previous world record of 403 seconds, also set by EAST in 2023. [...] The 1,000-second mark is considered a critical threshold in fusion research. Sustaining plasma for such extended durations is essential for demonstrating the feasibility of operating fusion reactors. This breakthrough, accomplished by the Institute of Plasma Physics under the CAS, signifies a major leap towards realizing the potential of fusion energy. [...] The success of EAST's recent experiment can be attributed to several key advancements. Researchers have made significant strides in improving the stability of the heating system, enhancing the accuracy of the control system, and refining the precision of the diagnostic systems. Warning: the source originates from China Daily, an English-language daily newspaper owned by the Central Propaganda Department of the Chinese Communist Party. It's rated "questionable" by Media Bias/Fact Check because of its association with the CCP.
China

DHS Terminates All Its Advisory Committees, Ending Its Investigation Into Chinese Telecom Hack (arstechnica.com) 144

An anonymous reader quotes a report from Ars Technica: The Department of Homeland Security has terminated all members of advisory committees, including one that has been investigating a major Chinese hack of large US telecom firms. "The Cyber Safety Review Board -- a Department of Homeland Security investigatory body stood up under a Biden-era cybersecurity executive order to probe major cybersecurity incidents -- has been cleared of non-government members as part of a DHS-wide push to cut costs under the Trump administration, according to three people familiar with the matter," NextGov/FCW reported yesterday.

A memo sent Monday by DHS Acting Secretary Benjamine Huffman said that in order to "eliminate[e] the misuse of resources and ensur[e] that DHS activities prioritize our national security, I am directing the termination of all current memberships on advisory committees within DHS, effective immediately. Future committee activities will be focused solely on advancing our critical mission to protect the homeland and support DHS's strategic priorities." The memo said advisory board members terminated this week "are welcome to reapply." The Cyber Safety Review Board's list of members included security experts from the private sector and lead cybersecurity officials from multiple government agencies.
"The CSRB was 'less than halfway' done with its Salt Typhoon investigation, according to a now-former member," wrote freelance cybersecurity reporter Eric Geller, who quoted an anonymous source as saying the Cyber Safety Review Board's review of Salt Typhoon is "dead." The former member was also quoted as saying, "There are still professional staff for the CSRB and I hope they will continue some of the work in the interim."

The Cyber Safety Review Board operates under (PDF) the DHS's Cybersecurity and Infrastructure Security Agency (CISA), notes Ars. The review board previously investigated a 2023 hack of Microsoft Exchange Online and more recently has been investigating how the Chinese hacking group called Salt Typhoon infiltrated major telecom providers such as Verizon and AT&T.
AI

Cutting-Edge Chinese 'Reasoning' Model Rivals OpenAI o1 55

An anonymous reader quotes a report from Ars Technica: On Monday, Chinese AI lab DeepSeek released its new R1 model family under an open MIT license, with its largest version containing 671 billion parameters. The company claims the model performs at levels comparable to OpenAI's o1 simulated reasoning (SR) model on several math and coding benchmarks. Alongside the release of the main DeepSeek-R1-Zero and DeepSeek-R1 models, DeepSeek published six smaller "DeepSeek-R1-Distill" versions ranging from 1.5 billion to 70 billion parameters. These distilled models are based on existing open source architectures like Qwen and Llama, trained using data generated from the full R1 model. The smallest version can run on a laptop, while the full model requires far more substantial computing resources.

The releases immediately caught the attention of the AI community because most existing open-weights models -- which can often be run and fine-tuned on local hardware -- have lagged behind proprietary models like OpenAI's o1 in so-called reasoning benchmarks. Having these capabilities available in an MIT-licensed model that anyone can study, modify, or use commercially potentially marks a shift in what's possible with publicly available AI models. "They are SO much fun to run, watching them think is hilarious," independent AI researcher Simon Willison told Ars in a text message. Willison tested one of the smaller models and described his experience in a post on his blog: "Each response starts with a ... pseudo-XML tag containing the chain of thought used to help generate the response," noting that even for simple prompts, the model produces extensive internal reasoning before output.
Although the benchmarks have yet to be independently verified, DeepSeek reports that R1 outperformed OpenAI's o1 on AIME (a mathematical reasoning test), MATH-500 (a collection of word problems), and SWE-bench Verified (a programming assessment tool).

TechCrunch notes that three Chinese labs -- DeepSeek, Alibaba, and Moonshot AI's Kimi, have released models that match o1's capabilities.

Slashdot Top Deals