Author Archives: Will Shanklin

YouTube is making tools to detect face and voice deepfakes

Posted on by

YouTube is developing new tools to protect artists and creators from the unauthorized use of their likenesses. The company said on Thursday that new tech to detect AI-generated content using a person’s face or singing voice is in the pipeline, with pilot programs starting early next year.

The upcoming face-detection tech will allegedly let people from various industries “detect and manage” content that uses an AI-generated depiction of their face. YouTube says it’s building the tools to allow creators, actors, musicians and athletes to find and choose what to do about videos that include a deepfake version of their likeness. The company hasn’t yet specified a release date for the face detection tools.

Meanwhile, the “synthetic-singing identification” tech will be part of Content ID, YouTube’s automated IP protection system. The company says the tool will let partners find and manage content that uses AI-generated versions of their singing voices.

“As AI evolves, we believe it should enhance human creativity, not replace it,” Amjad Hanif, YouTube’s vice president of creator products, wrote in a blog post. “We’re committed to working with our partners to ensure future advancements amplify their voices, and we’ll continue to develop guardrails to address concerns and achieve our common goals.”

This article originally appeared on Engadget at https://www.engadget.com/ai/youtube-is-making-tools-to-detect-face-and-voice-deepfakes-191536027.html?src=rss

Balatro is coming to mobile, and it’s going to ruin my life

Posted on by

Beware: The most addictive game of 2024 is coming to a device that’s always within reach. Balatro, a clever combination of poker, deck-building and roguelikes, will arrive on iOS — as a standalone purchase and Apple Arcade — and Android later this month. (That’s plenty of time to shop for a power bank or battery pack.)

Created by Canadian solo developer LocalThunk and published by Playstack, Balatro (styled as Balatro+ on Apple Arcade) is one of Engadget’s picks for best PC games. It's one of the great indie gaming success stories of 2024.

Unlike poker, Balatro’s spin includes magical tarot cards and unique Jokers that can dramatically alter your hands. For example, one might increase the pool of cards to build your hand from, and another could raise your score if your hand hits a particular sequence. Depending on which Joker you draw, you may be forced to rethink your entire strategy. (But play them strategically because you’re limited to five per hand.)

Its presentation adds to the fun. It includes a synthwave soundtrack, trippy animations and a delightfully retro CRT filter.

If you don’t know much about poker, don’t sweat it. First, poker is only one element of the game — and, even then, it encourages you to break its rules. Balatro also includes tutorials and help pop-ups to make learning the unique deck game (and throwing away all your free time) as easy as possible.

Balatro launched in February on PC, PlayStation, Xbox and Switch (and on macOS in March). It arrives on Android and iOS on September 26. It will be a $10 standalone purchase on either mobile platform or free on Apple devices for Apple Arcade subscribers. Those in the Apple ecosystem can also play it on iPad, Mac, Apple TV and Vision Pro.

Screenshot from NBA 2K25 Arcade Edition. Recreation of a vintage game with Vince Carter (Raptors era) rising up for a slam as Chris Bosh runs along the left wing.
2K Sports / Apple

NBA 2K25 Arcade Edition, which continues the franchise’s mobile-friendly spinoffs for Apple devices, is also coming soon to Apple Arcade. This year’s version adds The Neighborhood, the open-world, skill-building environment first seen in the standard version of NBA 2K18. It also adds a revamped version of Greatest Mode, where you can relive iconic moments from NBA legends, past and present. 2K Sports promises new Great Mode players will be added throughout the season.

NBA 2K Arcade Edition arrives on Apple Arcade on October 3.

This article originally appeared on Engadget at https://www.engadget.com/gaming/balatro-is-coming-to-mobile-and-its-going-to-ruin-my-life-173039968.html?src=rss

Google’s Gemini-powered photo search arrives in early access

Posted on by

Google’s AI-powered Photos upgrades are beginning to trickle in. Ask Photos, the Gemini-powered chatbot that lets you get ultra-specific and conversational with your photo searches, is launching in early access for select users in the US. In addition, the improved search for more descriptive Google Photos queries begins rolling out today for all English-speaking users.

The upgraded search in Google Photos lets you use more descriptive queries. For example, while you could have searched for “lake” before, you can now enter “kayaking on a lake surrounded by mountains.” Or, instead of merely searching for your friend Alice, you can go with “Alice and me laughing.” The idea is to make it easier to narrow things down as our cloud-based photo libraries grow.

Ask Photos, the Google Photos chatbot the company revealed at I/O in May, takes that further. Powered by Gemini, it adds a new tab at the bottom of the Photos app that lets you ask about anything in your library using natural language.

Google provided examples like “Show me the best photo from each national park I’ve visited,” which uses location data to scour your park photos and some subjective robot judgment to determine a favorite. Other examples the company provided include “What did we eat at the hotel in Stanley?” and “Where did we camp last time we went to Yosemite?”

Like other chatbot features, Ask Photos can respond to follow-up prompts. So, if it misses the mark the first time, you can ask it to tweak its parameters and give it another go.

Google says your Photos data will never be used for advertising. Although humans may review queries, they’ll be disconnected from your Google account, so the reviewers won’t know who typed the input. Real people won’t review Ask Google’s answers, including photos or videos, unless you provide feedback or (only in rare cases, according to the company) to address abuse.

If you’re in the US, you can sign up for the waitlist to try to get early access to Ask Photos starting today. Meanwhile, Google Photos’ more descriptive search powers are now beginning to roll out for English-speaking users on Android and iOS.

This article originally appeared on Engadget at https://www.engadget.com/ai/googles-gemini-powered-photo-search-arrives-in-early-access-160041679.html?src=rss

Verizon is reportedly near a deal to buy broadband provider Frontier Communications

Posted on by

Verizon is reportedly near a deal to buy fiber provider Frontier Communications. On Wednesday, The Wall Street Journal said that an announcement could come as early as this week, provided discussions don’t “hit any last-minute snags.”

Frontier has a market value of over $7 billion and provides broadband to around three million locations in 25 states. The company would help Verizon boost its Fios fiber network and better compete with AT&T. The carrier has seen slowing wireless revenue and views fiber investment as a growth area. Acquiring companies with existing infrastructure, like Frontier, is potentially less expensive and time-consuming than rolling out its own network.

Based in Dallas, Frontier is currently upgrading its copper landline system to fiber — enabling it to offer a 5Gbps symmetrical plan. The company filed for Chapter 11 bankruptcy in 2020. It pivoted to a “leaner business,” as the WSJ describes, before running into concerns that it would run out of money before it finishes its current upgrades.

The FTC sued the company in 2021 for misrepresenting its speeds. Under a 2022 settlement, Frontier was required to stop lying about its internet performance, dole out over $8.5 million and install fiber service in 60,000 California homes over four years.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/verizon-is-reportedly-near-a-deal-to-buy-broadband-provider-frontier-communications-210317747.html?src=rss

Volvo scales back its EV goals, will no longer be fully electric by 2030

Posted on by

Over three years after saying it would sell only electric vehicles by 2030, Volvo has lowered its EV ambitions. The automaker now says it will aim for 90 to 100 percent electrified vehicles (including full EVs and plug-in hybrids) by the decade’s end, with the remaining 0 to 10 percent being mild hybrids. Volvo chalked up its revised ambitions to “changing market conditions and customer demands.”

Volvo says it’s still committed to long-term electrification. The automaker has launched five fully electric models since laying out its (now aborted) 2030 goal three years ago: the EX40, EC40, EX30, EM90 and EX90.

The company cites the slower-than-expected rollout of EV charging infrastructure as one factor in its decision. Despite the passage of President Biden’s Bipartisan Infrastructure Law in 2021, which allocated $7.5 billion to support the creation of 500,000 EV charging stations, only seven stations in four states had been built as of March. Reasons for the slow rollout allegedly include a lack of experience in the state transportation agencies in charge of execution and various government requirements (submitting plans, soliciting bids, awarding funds).

The Biden Administration said earlier this year it still expects the US to reach 500,000 charging stations by 2026.

Volvo also cited “additional uncertainties created by recent tariffs on EVs in various markets.” That likely refers to the hit the automaker is taking from manufacturing some models in China. Earlier this year, the White House announced new levies on EVs made in China and batteries sourced from China. (Volvo’s parent company, Volvo Car AB, is majority-owned by China’s Geely Holding.) Forbes reported in May that the China-made EX30, expected to start at around $37,000, would be pushed to over $50,000 after tariffs.

The automaker adjusted its CO2 reduction expectations alongside the tweaked timeline. It now says it aims for 65 to 75 percent reduced per-car emissions (compared to a 2018 baseline) by 2030; its previous goal was a hard 75 percent. In addition, it also changed its previous 40-percent per-car reduction (also compared to 2018) by 2025; that goal is now a 30 to 35 percent drop.

“We are resolute in our belief that our future is electric,” Jim Rowan, Volvo Cars CEO, wrote in a press release. “An electric car provides a superior driving experience and increases possibilities for using advanced technologies that improve the overall customer experience. However, it is clear that the transition to electrification will not be linear, and customers and markets are moving at different speeds of adoption. We are pragmatic and flexible, while retaining an industry-leading position on electrification and sustainability.”

This article originally appeared on Engadget at https://www.engadget.com/transportation/evs/volvo-scales-back-its-ev-goals-will-no-longer-be-fully-electric-by-2030-201059287.html?src=rss

Cheaper Copilot+ PCs are coming with Qualcomm’s 8-core Snapdragon X Plus chip

Posted on by

Qualcomm is moving to make AI PCs more affordable. Following the company’s 12-core Snapdragon X Elite and 10-core Snapdragon X Plus, it unveiled a toned-down eight-core version of the Snapdragon X Plus on Wednesday. The chip includes the same Hexagon neural processing unit (NPU) from the higher-end variants, capable of 45 trillion operations per second (TOPS) for powerful on-device AI.

The 4nm AI-focused chip has a custom Qualcomm Orion CPU built for “mainstream” (i.e., cheaper) Copilot+ PCs. Its eight cores can reach speeds of up to 3.2GHz, with single-core performance at up to 3.4GHz. Qualcomm says it enables days-long battery life in laptops.

The chip includes an integrated Adreno GPU, which supports up to three 4K 60Hz monitors or two 5K 60Hz displays. It supports an internal display of up to UHD 120Hz with HDR10.

The chart below shows how the Snapdragon X Plus 8-core’s specs compare to other AI chips in the line:

A chart showing different Snapdragon X AI chips
Qualcomm

“Copilot+ PCs, powered exclusively today by Snapdragon X Series platforms, launched the new generation in personal computing, made possible by our groundbreaking NPU,” Qualcomm President and CEO Cristiano Amon wrote in a press release. “We are now bringing these transformative AI experiences, along with best-in-class performance and unprecedented battery life, to more users worldwide with Snapdragon X Plus 8-core. We’re proud to be working with our global OEM partners to restore performance leadership to the Windows ecosystem.”

The first PCs with the 8-core Snapdragon X Plus include laptops from Acer, Dell, HP, Lenovo and others. They’ll be available starting today.

This article originally appeared on Engadget at https://www.engadget.com/computing/cheaper-copilot-pcs-are-coming-with-qualcomms-8-core-snapdragon-x-plus-chip-110013598.html?src=rss

Copilot+ features are coming in November to AI PCs powered by Intel and AMD’s latest chips

Posted on by

Qualcomm’s exclusivity period on Copilot+ PCs is winding down. Microsoft confirmed on Tuesday that Intel’s new 200V processors and AMD’s Ryzen AI 300 series chips will add Copilot+ AI capabilities beginning in November.

Copilot+ PCs include features like Live Captions (real-time subtitle generation, including translations), Cocreator in Paint (prompt-based image generation), Windows Studio Effects image editing (background blurring, eye contact adjustment and auto-framing) and AI tools in Photos. Of particular interest to gamers is Auto Super Resolution, an Nvidia DLSS competitor that upscales graphical resolution and refresh rates in real time without stunting performance.

The AI PCs will also eventually include Recall, Microsoft’s searchable timeline of PC activity. This feature was delayed to enhance security after an initial blowback. (Who’d have thought a history of everything you do on your PC might need to be locked down as tightly as possible?) The company said the revised Recall would start rolling out to beta testers in October.

Chart showing
Intel

Intel’s 200V series processors, revealed today, include a powerful neural processing unit (NPU) that supports up to 48 TOPS (tera operations per second) for locally processed AI models and tools. With up to 32GB of onboard memory, the 200V is “the most efficient x86 processor ever,” according to Intel, with 50 percent lower on-package power consumption.

Microsoft’s Windows and devices lead, Pavan Davuluri, confirmed that Intel’s new chips will support Copilot+. “All designs featuring Intel Core Ultra 200V series processors and running the latest version of Windows are eligible to receive Copilot+ PC features as a free update starting in November,” Davuluri said onstage at Intel’s IFA launch event in Germany.

Meanwhile, according to a Windows blog post, AMD’s Ryzen AI 300 series chips, revealed earlier this summer, will also receive Copilot+ features in November. The NPUs in AMD’s chips can reach up to 50 TOPS for AI performance and have 16 percent faster overall performance than their predecessors.

The first Copilot+ PCs arrived in June, powered by Qualcomm’s Snapdragon X Elite chip. The initial batch of Arm-based PCs include laptops and 2-in-1s from Microsoft, Acer, HP, Lenovo, Samsung, Asus and Dell.

This article originally appeared on Engadget at https://www.engadget.com/ai/copilot-features-are-coming-to-ai-pcs-powered-by-intel-and-amds-latest-chips-190707475.html?src=rss

MLB’s virtual ballpark returns for four regular-season games in September

Posted on by

Major League Baseball’s virtual ballpark is back. Like a metaverse experience for traditional (non-VR) devices, it lets you watch actual games as they're happening in real time, albeit recreated in a 3D environment. MLB will host interactive watch parties in the environment for select games each Wednesday in September.

Like a baseball-centric take on Second Life (for the old folks in the back), it includes 3D avatars corresponding to players’ movements. The plays and athletes’ precise positions are tracked using the same Sony Hawk-Eye cameras used for the league’s Statcast analytics system. The experience sounds tailor-made for headsets like the Vision Pro and Meta Quest, but it’s limited to traditional screens for now. Improbable, a London-based company known for metaverse experiences, created the tech.

There’s also an audio element, as you can hear the play-by-play and game sounds and chat with other fans in spatial audio. New for this season is a party system that lets you talk directly with your friends. The league is also bringing back a virtual scavenger hunt to keep you interested in case the game is a bore.

MLB’s virtual ballpark debuted during the 2023 season, first for an All-Star exhibition and then for a late regular-season game and a Postseason matchup. This season’s virtual lineup kicks off on Wednesday, with three more games following throughout September:

  • Wednesday, September 4, 6:50PM ET - Tamba Bay Rays vs. Minnesota Twins

  • Wednesday, September 11, 7:45PM ET - Reds vs. Cardinals

  • Wednesday, September 18, 6:35PM ET - Giants vs. Orioles

  • Wednesday, September 25, 6:40PM ET - Rays vs. Tigers

It isn’t clear who this experience was made for, but hey, at least it’s free. You can log into the virtual ballpark using any modern device with a web browser at MLB’s virtual ballpark website. You’ll need to create or log into an MLB account before they let you past the virtual turnstile.

This article originally appeared on Engadget at https://www.engadget.com/entertainment/mlbs-virtual-ballpark-returns-for-four-regular-season-games-in-september-171533897.html?src=rss

Google is rolling out more election-related safeguards in YouTube, search and AI

Posted on by

As the US speeds toward one of the most consequential elections in its 248-year history, Google is rolling out safeguards to ensure users get reliable information. In addition to the measures it announced late last year, the company said on Friday that it’s adding election-related guardrails to YouTube, Search, Google Play and AI products.

YouTube will add information panels above the search results for at least some federal election candidates. The modules, likely similar to those you see when searching the web for prominent figures, will include the candidates’ basic details like their political party and a link to Google Search for more info. The company says the panels may also include a link to the person’s official website (or other channel). As Election Day (November 5) approaches, YouTube’s homepage will also show reminders on where and how to vote.

Google Search will include aggregated voter registration resources from state election offices for all users. Google is sourcing that data through a partnership with Democracy Works, a nonpartisan nonprofit that works with various companies and organizations “to help voters whenever and wherever they need it.”

Meanwhile, the Google Play Store will add a new badge that indicates an app is from an official government agency. The company outlines its requirements for apps that “communicate government information” in a developer help document. Approved applications that have submitted the required forms are eligible for the “official endorsement signified by a clear visual treatment on the Play Store.”

As for generative AI, which can be prone to hallucinations that would make Jerry Garcia blush, Google is expanding its election-related restrictions, which were announced late last year. They’ll include disclosures for ads created or generated using AI, content labels for generated content and embedded SynthID digital watermarking for AI-made text, audio, images and video. Initially described as being for Gemini (apps and on the web), the election guardrails will apply to Search AI Overviews, YouTube AI-generated summaries for Live Chat, Gems (custom chatbots with user-created instructions) and Gemini image generation.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/google-is-rolling-out-more-election-related-safeguards-in-youtube-search-and-ai-190422568.html?src=rss

OpenAI and Anthropic agree to share their models with the US AI Safety Institute

Posted on by

OpenAI and Anthropic have agreed to share AI models — before and after release — with the US AI Safety Institute. The agency, established through an executive order by President Biden in 2023, will offer safety feedback to the companies to improve their models. OpenAI CEO Sam Altman hinted at the agreement earlier this month.

The US AI Safety Institute didn’t mention other companies tackling AI. But in a statement to Engadget, a Google spokesperson told Engadget the company is in discussions with the agency and will share more info when it’s available. This week, Google began rolling out updated chatbot and image generator models for Gemini.

“Safety is essential to fueling breakthrough technological innovation. With these agreements in place, we look forward to beginning our technical collaborations with Anthropic and OpenAI to advance the science of AI safety,” Elizabeth Kelly, director of the US AI Safety Institute, wrote in a statement. “These agreements are just the start, but they are an important milestone as we work to help responsibly steward the future of AI.”

The US AI Safety Institute is part of the National Institute of Standards and Technology (NIST). It creates and publishes guidelines, benchmark tests and best practices for testing and evaluating potentially dangerous AI systems. “Just as AI has the potential to do profound good, it also has the potential to cause profound harm, from AI-enabled cyber-attacks at a scale beyond anything we have seen before to AI-formulated bioweapons that could endanger the lives of millions,” Vice President Kamala Harris said in late 2023 after the agency was established.

The first-of-its-kind agreement is through a (formal but non-binding) Memorandum of Understanding. The agency will receive access to each company’s “major new models” ahead of and following their public release. The agency describes the agreements as collaborative, risk-mitigating research that will evaluate capabilities and safety. The US AI Safety Institute will also collaborate with the UK AI Safety Institute.

It comes as federal and state regulators try to establish AI guardrails while the rapidly advancing technology is still nascent. On Wednesday, the California state assembly approved an AI safety bill (SB 10147) that mandates safety testing for AI models that cost more than $100 million to develop or require a set amount of computing power. The bill requires AI companies to have kill switches that can shut down the models if they become “unwieldy or uncontrollable.”

Unlike the non-binding agreement with the federal government, the California bill would have some teeth for enforcement. It gives the state’s attorney general license to sue if AI developers don’t comply, especially during threat-level events. However, it still requires one more process vote — and the signature of Governor Gavin Newsom, who will have until September 30 to decide whether to give it the green light.

Update, August 29, 2024, 4:53 PM ET: This story has been updated to add a response from a Google spokesperson.

This article originally appeared on Engadget at https://www.engadget.com/ai/openai-and-anthropic-agree-to-share-their-models-with-the-us-ai-safety-institute-191440093.html?src=rss