Tencent Touts Open-Source AI Models to Turn Text into 3D Visuals
Tencent Holdings Ltd. released new AI services that turn text or images into 3D visuals and graphics, the latest in a series of products to emerge from big tech firms since DeepSeek galvanized Chinese and US artificial intelligence development.
Tencent's five new 3D-content generators are built atop its Hunyuan3D-2.0 model, all of which it intends to open-source to users, the company said in a statement. They will help power an upgraded version of Tencent's proprietary 3D engine for games and other content.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
27 minutes ago
- Yahoo
Stargate UAE AI Deal Stalls Over U.S. Chip Rules
AI behemoth Stargate UAE has hit a roadblock because U.S. officials are still hashing out security terms for exporting advanced chips. Nvidia (NASDAQ:NVDA), Oracle (NYSE:ORCL), OpenAI, Cisco Systems (NASDAQ:CSCO), SoftBank Group (SFTBY) and UAE's G42 teamed up last month after President Trump's Middle East stop to launch what could become a 5-gigawatt AI data-center powerhouse. Warning! GuruFocus has detected 4 Warning Signs with NVDA. The first 200-megawatt chunk was supposed to go live in 2026, operated by OpenAI and Oracle, but nobody's signed off on the chip-export rules yet, so the deal remains on ice. Although G42 says the project builds on the new U.S.-UAE AI Acceleration Partnership, which aims to foster safe, responsible AI, Washington still worries about the UAE's China tiesrecalling how the UAE let Huawei (now under U.S. fire) roll out 5G during Trump's first term. OpenAI and G42 both have Microsoft (NASDAQ:MSFT) backing, but even that doesn't erase concerns that Chinese hardware or talent could leak U.S. tech to adversaries. Because of those unresolved security questionslike banning Chinese gear on site or limiting who can work therethe Commerce Department, Nvidia, OpenAI, G42, Microsoft, Oracle and SoftBank all declined to say when any approval might come. And adding drama, Elon Musk recently hinted he might torpedo the whole thing unless xAI gets a seat at the table. Investors should care because if the UAE pushes back on U.S. curbsinsisting on looser rulesthe entire Stargate plan could stall, undermining a high-profile attempt to export American AI clout. Watch for any new U.S. export guidelines and whether Abu Dhabi caves to get those cutting-edge chips; otherwise, this megaproject could sit stalled well past 2026. This article first appeared on GuruFocus. Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
CNBC
33 minutes ago
- CNBC
Trump announces U.S.-China trade talks in London next week
Treasury Secretary Scott Bessent and two other Trump administration officials will meet with their Chinese counterparts in London on Monday for renewed trade talks, President Donald Trump said. Bessent, who has been leading the administration's efforts to craft a deal with Beijing, will be joined by Commerce Secretary Howard Lutnick and U.S. Trade Representative Jamieson Greer, Trump said. "The meeting should go very well," the president wrote on Truth Social on Friday afternoon. "Thank you for your attention to this matter!" CNBC has contacted the Chinese Embassy in Washington for more information on Trump's announcement. A spokesperson for the embassy did not immediately reply to a request for comment. This is breaking news. Please refresh for updates.
The Hill
42 minutes ago
- The Hill
Mission possible: An alternative to facial recognition technology
For decades, Hollywood has presented audiences with futuristic disguises that were once thought only possible in science fiction. Silicone masks, fake contact lenses, and 3D printed biometrics are staples of popular spy movies like the 'Mission: Impossible' franchise. But these forms of 'spyware,' once found only on the silver screen, are, in fact, a reality. The advent of the internet and facial recognition technology has turned disguise work into a matter of national security. Just look at how our adversaries abuse facial recognition technology. Whether it is the Social Credit System or cameras lining public streets to monitor dissent against the Chinese state or oppress minority groups, China's surveillance state is built on facial recognition technology — some of the most sophisticated in the world, due to the amount of data it can access through measures such as its National Security Law. The Chinese Communist Party's monitoring system can essentially control the life of any individual across its regions, freeze payments and track purchases anywhere in the country. With China as an example of how not to use facial recognition technology, the U.S. should be clear-eyed about the vulnerabilities and potential abuses posed by these increasing outmoded forms of biometric security. Advancements in artificial intelligence, deepfakes and three-dimensional printing are successfully tricking facial recognition tools, which should affirm that we cannot continue to rely on them to protect locations critical to national security. Threat actors have developed a variety of tactics to spoof facial recognition software. Some are known as 'replay attacks' and occur when a video is presented to a facial biometric system by an actor other than the intended user. Static photos are another form of attack with the same intention. Currently, the success rate for bypassing facial recognition technologies with these methods is 98 percent and 96 percent respectively — a staggering statistic. Europol even recently noted how artificial intelligence is successfully compromising phones, issuing a warning about 'increased use of artificial fingerprints, deepfake media, and voice cloning to bypass security protocols.' Facial recognition technology also presents challenges with accurate identification. While cheap or generic silicone masks perform poorly in fooling individual biometric devices, they prove effective in avoiding facial recognition technology in a crowd. Let us also not forget that some older versions of facial recognition technology struggled to distinguish between people with darker skin. Knowing this challenge, we must seriously question efforts to use facial recognition to verify voter identity before casting a ballot — one of America's foundational processes. There are three things we must do as we move away from facial recognition technology. First, to protect the privacy of Americans, and until new technology is put into place, the U.S. should begin by exploring implementation of proven biometric security tools on a solely voluntary basis. For example, the Transportation Security Administration uses facial verification technology, and permits individuals to opt out. Second, consequences must be imposed when unsecure technology is developed, or adversaries cross the line. While the Committee on Homeland Security is currently undertaking the challenge to change these economic models in cybersecurity, we have an opportunity to get ahead of them now by pursuing more secure and accurate biometric security tools. We cannot become overly dependent on fallible technology — the risks are simply too high. Finally, while we seek alternatives to facial recognition technology that ensure U.S. law enforcement entities have the best tools to protect us, the U.S. must clearly call out China for its abuse of facial recognition technology. The Chinese Communist Party's use of facial recognition technology to control its citizenry is unacceptable and should concern all Americans. The U.S. must address the risks of facial recognition technology head on to protect the liberties we cherish. It is time we work closely with our innovators to champion biometric solutions that are secure, reliable and aligned with American values. Mark Green, M.D., represents Tennessee's 7th Congressional District and is chairman of the House Homeland Security Committee.
