Apple pulled the latest short film in its The Underdogs: OOO (Out of Office) series set in Thailand. The tech giant scrubbed it over complaints about stereotypical portrayals of Thailand and its people in certain scenes.
reports that Apple issued an apology to the people of Thailand for the fifth film in its Underdogs series. The ad series features a group of travel weary office workers navigating the world using Apple’s various products.
Several viewers posted comments criticizing the film’s use of a sepia filter to make Thailand seem underdeveloped. The comments also called out the costuming and scenery decisions in its airport scene using outdated representations of Thailand’s citizens.
Sattra Sripan, the spokesman for the Thai House of Representatives’ committee on tourism, called for a boycott over the ad.
“Thai people are deeply unhappy with the advertisement,” Sripan said in a statement. “I encourage Thai people to stop using Apple products and change to other brands.”
Apple issued an apology for the ad shortly after pulling it off of YouTube. Lawmakers have also invited Apple representatives to visit with them to discuss the ads and how they portray Thailand on film.
“Our intent was to celebrate the country’s optimism and culture, and we apologize for not fully capturing the vibrancy of Thailand today,” the statement read.
This is the second time this year that Apple has apologized for a commercial. that it told AdAge “missed the mark” for its new thin iPad Pro. The commercial features a giant pneumatic press crushing a large collection of items used in or to represent creative endeavors such musical instruments, paints, a generic arcade cabinet, and camera equipment. The steel crusher smooshes everything flat and lifts up to reveal an intact iPad sitting on the lower steel block that a voiceover describes as “the most powerful iPad ever is also the thinnest.”
Artists, musicians and other creators took offense to the ad’s implied tone that generative AI would replace human artistic endeavors. Apple vowed not to air the ad on TV but it’s still on its YouTube page with the comments section disabled.
When most tech companies are challenged with a lawsuit, the expected defense is to deny wrongdoing. To give a reasonable explanation of why the business’ actions were not breaking any laws. Music AI startups Udio and Suno have gone for a different approach: admit to doing exactly what you were sued for.
Udio and Suno were sued in June, with music labels Universal Music Group, Warner Music Group and Sony Music Group claiming they trained their AI models by scraping copyrighted materials from the Internet. In a court filing today, Suno acknowledged that its neural networks do in fact scrape copyrighted material: “It is no secret that the tens of millions of recordings that Suno’s model was trained on presumably included recordings whose rights are owned by the Plaintiffs in this case.” And that’s because its training data “includes essentially all music files of reasonable quality that are accessible on the open internet,” which likely include millions of illegal copies of songs.
But the company is taking the line that its scraping falls under the umbrella of fair use. “It is fair use under copyright law to make a copy of a protected work as part of a back-end technological process, invisible to the public, in the service of creating an ultimately non-infringing new product,” the statement reads. Its argument seems to be that since the AI-generated tracks it creates don’t include samples, illegally obtaining all of those tracks to train the AI model isn’t a problem.
Calling the defendants’ actions “evading and misleading,” the RIAA, which initiated the lawsuit, had an unsurprisingly harsh response to the filing. “Their industrial scale infringement does not qualify as ‘fair use’. There’s nothing fair about stealing an artist’s life’s work, extracting its core value, and repackaging it to compete directly with the originals,” a spokesperson for the organization said. “Defendants had a ready lawful path to bring their products and tools to the market – obtain consent before using their work, as many of their competitors already have. That unfair competition is directly at issue in these cases.”
Whatever the next phase of this litigation entails, prepare your popcorn. It should be wild.
A new government program is trying to encourage Internet service providers (ISPs) to offer lower rates for lower income customers by distributing federal funds through states. The only problem is the ISPs don’t want to offer the proposed rates.
obtained a letter sent to US Commerce Secretary Gina Raimondo signed by more than 30 broadband industry trade groups like ACA Connects and the Fiber Broadband Association as well as several state based organizations. The letter raises “both a sense of alarm and urgency” about their ability to participate in the Broadband Equity, Access and Deployment (BEAD) program. The newly formed BEAD program provides over $42 billion in federal funds to “expand high-speed internet access by funding planning, infrastructure, deployment and adoption programs” in states across the country, according to the (NTIA).
The money first goes to the NTIA and then it’s distributed to states after they obtain approval from the NTIA by presenting a low-cost broadband Internet option. The ISP industries’ letter claims a fixed rate of $30 per month for high speed Internet access is “completely unmoored from the economic realities of deploying and operating networks in the highest-cost, hardest-to-reach areas.”
The letter urges the NTIA to revise the low-cost service option rate proposed or approved so far. have completed all of the BEAD program’s phases.
Americans pay an average of $89 a month for Internet access. New Jersey has the highest average bill at $126 per month, according to a survey conducted by . A 2021 study from the found that 57 percent of households with an annual salary of $30,000 or less have a broadband connection.
Three US Senators introduced a bill that aims to rein in the rise and use of AI generated content and deepfakes by protecting the work of artists, songwriters and journalists.
The COPIED ACT would, if enacted, create transparency standards through the National Institutes of Standards and Technology (NIST) to set guidelines for “content provenance information, watermarking, and synthetic content detection,” according to the press release.
The bill would also prohibit the unauthorized use of creative or journalistic content to train AI models or created AI content. The Federal Trade Commission and state attorneys general would also gain the authority to enforce these guidelines and individuals who had their legally created content used by AI to create new content without their consent or proper compensation would also have the right to take those companies or entities to court.
The bill would even expand the prohibition of tampering or removing content provenance information by internet platforms, search engines and social media companies.
A slew of content and journalism advocacy groups are already voicing their support for the COPIED Act to become law. They include groups like SAG-AFTRA, the Recording Industry Association of America, the National Association of Broadcasters, the Songwriters Guild of America and the National Newspaper Association.
This is not the Senate’s first attempt to create guidelines and laws for the rising use of AI content and it certainly won’t be the last. In April, Rep. Adam Schiff (D-Calif.) submitted a bill called the Generative AI Copyright Disclosure Act that would force AI companies to list their copyrighted sources in their datasets. The bill has not moved out of the House Committee on the Judiciary since its introduction, according to Senate records.
Most subscription sites use “dark patterns” to influence customer behavior around subscriptions and personal data, according to a pair of new reports from global consumer protection groups. Dark patterns are “practices commonly found in online user interfaces [that] steer, deceive, coerce or manipulate consumers into making choices that often are not in their best interests.” The international research efforts were conducted by the International Consumer Protection and Enforcement Network (ICPEN) and the Global Privacy Enforcement Network (GPEN).
The ICPEN conducted the of 642 websites and mobile apps with a subscription component. The assessment revealed one dark pattern in use at almost 76 percent of the platforms, and multiple dark patterns at play in almost 68 percent of them. One of the most common dark patterns discovered was sneaking, where a company makes potentially negative information difficult to find. ICPEN said 81 percent of the platforms with automatic subscription renewal kept the ability for a buyer to turn off auto-renewal out of the purchase flow. Other dark patterns for subscription services included interface interference, where desirable actions are easier to perform, and forced action, where customers have to provide information to access a particular function.
The companion from GPEN examined dark patterns that could encourage users to compromise their privacy. In this review, nearly all of the more than 1,000 websites and apps surveyed used a deceptive design practice. More than 89 percent of them used complex and confusing language in their privacy policies. Interface interference was another key offender here, with 57 percent of the platforms making the least protective privacy option the easiest to choose and 42 percent using emotionally charged language that could influence users.
Even the most savvy of us can be influenced by these subtle cues to make suboptimal decisions. Those decisions might be innocuous ones, like forgetting that you’ve set a service to auto-renew, or they might put you at risk by encouraging you to reveal more personal information than needed. The reports didn’t specify whether the dark patterns were used in illicit or illegal ways, only that they were present. The dual release is a stark reminder that digital literacy is an essential skill.
OpenAI seems to make headlines every day and this time it’s for a double dose of security concerns. The first issue centers on the Mac app for ChatGPT, while the second hints at broader concerns about how the company is handling its cybersecurity.
Earlier this week, engineer and Swift developer Pedro José Pereira Vieito the Mac ChatGPT app and found that it was storing user conversations locally in plain text rather than encrypting them. The app is only available from OpenAI’s website, and since it’s not available on the App Store, it doesn’t have to follow Apple’s sandboxing requirements. Vieito’s work was then covered by and after the exploit attracted attention, OpenAI released an update that added encryption to locally stored chats.
For the non-developers out there, sandboxing is a security practice that keeps potential vulnerabilities and failures from spreading from one application to others on a machine. And for non-security experts, storing local files in plain text means potentially sensitive data can be easily viewed by other apps or malware.
The second issue occurred in 2023 with consequences that have had a ripple effect continuing today. Last spring, a hacker was able to obtain information about OpenAI after illicitly accessing the company’s internal messaging systems. reported that OpenAI technical program manager Leopold Aschenbrenner raised security concerns with the company’s board of directors, arguing that the hack implied internal vulnerabilities that foreign adversaries could take advantage of.
Aschenbrenner now says he was fired for disclosing information about OpenAI and for surfacing concerns about the company’s security. A representative from OpenAI told The Times that “while we share his commitment to building safe A.G.I., we disagree with many of the claims he has since made about our work” and added that his exit was not the result of whistleblowing.
App vulnerabilities are something that every tech company has experienced. Breaches by hackers are also depressingly common, as are contentious relationships between whistleblowers and their former employers. However, between how broadly ChatGPT has been adopted into services and how chaotic the company’s , and have been, these recent issues are beginning to paint a more worrying picture about whether OpenAI can manage its data.
A state-owned power company is splashing out 80 billion yuan ($11 billion) on an energy base that will generate electricity from , and sources. China Three Gorges Renewables Group, a subsidiary of the country’s largest hydropower company, plans to build a plant with a 16-gigawatt capacity and a five-gigawatt storage facility, reports.
This is part of China’s aim to build 455 gigawatts worth of renewable energy projects in the desert by 2030. This plant is being constructed in Inner Mongolia, which will get 135 gigawatts of the total planned output.
The China Three Gorges Corporation is looking to diversify its energy sources as building large hydro dams is becoming less feasible. According to Three Gorges, wind and solar generation from the plant will depend on grid accessibility. The coal plant is set to start operations in three years.
It’s somewhat disappointing that the new plant will have a coal power element, though it’s not fully surprising given the way China has bristled at renewable energy commitments during . As Bloomberg notes, China has been struggling to put all of its clean energy into the power grid. It often relies on coal when renewable sources like solar and wind aren’t available.
The US has banned companies like Nvidia from selling their most advanced AI chips to China since 2022. But if loopholes exist, profit-hungry corporations will find and exploit them. The Informationpublished a bombshell report on Thursday detailing how Oracle allows TikTok owner ByteDance to rent Nvidia’s most advanced chips to train AI models on US soil.
ByteDance, which many US lawmakers believe has direct ties to the Chinese government, is reportedly renting US-based servers containing Nvidia’s coveted H100 chips from US cloud computing company Oracle to train AI models. The practice, which runs against the spirit of the US government’s chip regulations, is technically allowed because Oracle is merely renting out the chips on American soil, not selling them to companies in China.
The US government has cracked down on exporting the chips to China as an extension of the tensions between the two nations. The Biden Administration fears the nation could use advanced AI for military or surveillance purposes or to gain an economic upper hand. The US government passed bipartisan legislation in April that will force ByteDance to either sell its US operations or face a ban. But ByteDance still has until early next year to close a deal, and it’s suing the US government, which could delay enforcement.
Although ByteDance is training its models in the US, “it could be difficult to stop them from sending the models they produced back to their headquarters in China,” according to US-based cloud providers and a former Nvidia employee who spoke to The Information. Quite the loophole, indeed.
ByteDance’s Project Texas initiative, which the company claims siloes off TikTok’s US operations from its Chinese leadership to allay US fears, is at the heart of the arrangement. However, former ByteDance employees have described Project Texas as “largely cosmetic,” as they claim the company’s US wing regularly works closely with its Beijing-based leadership.
ByteDance isn’t the only Chinese company looking to game the rules. The Information says Alibaba and Tencent are discussing similar arrangements to gain access to the sought-after chips. Those deals could be harder to squash because they have their own US-based data centers and wouldn’t have to rent servers from American companies.
US cloud computing company Oracle reportedly enables ByteDance’s training of AI models in the US. (Oracle)
Not every company has been as willing as Oracle to skirt the law’s intent. “Two small American cloud providers” reportedly turned down offers to rent servers with Nvidia’s H100 chips to ByteDance and China Telecom because “they seemed to go against the spirit of U.S. chip restrictions.” However, Oracle, cofounded by American businessman Larry Ellison and run by current CEO Safra Catz, apparently found the opportunity for profit through technically legal workarounds too tempting to pass up.
The US Commerce Department, the bureau that could close the loophole, may already be aware of the practices. Earlier this year, the department proposed a rule that would require US cloud providers to verify foreign customers’ identities and notify the US if any of them were training AI models that “could be used in malicious cyber-enabled activity.” However, the Commerce Department recently said most cloud providers disapproved of the proposal, claiming “the burden of additional requirements might outweigh the intended benefit.” In the meantime, the proposed rule, which could theoretically plug the loophole, remains in limbo.
But even if the US manages to shut down that exploit, The Information says it wouldn’t cover Chinese cloud providers like Tencent and Alibaba from buying Nvidia’s chips and using them to train AI models in their own US-based data centers. The Commerce Department will have its hands full figuring this one out as business and defense interests wrestle for control.
Newly proposed Congressional legislation would require the US to conduct security reviews for connected vehicles built by automakers from China and “other countries of concern.” Rep. Elissa Slotkin (D-MI), a former CIA analyst and Pentagon official who has championed the issue, introduced the bill on Wednesday.
If passed by Congress (a tall order these days), the Connected Vehicle National Security Review Act would establish a formal review process for connected autos from Chinese companies. It would also allow the Department of Commerce to limit or ban these cars and other vehicles before they reach US consumers.
“Today’s vehicles are more sophisticated than ever, carrying cameras, radars and other sophisticated sensors, plus the ability to process, transmit and store the data they gather from the United States,” said Slotkin. “If allowed into our markets, Chinese connected vehicles offer the Chinese government a treasure trove of valuable intelligence on the United States, including the potential to collect information on our military bases, critical infrastructure like the power grid and traffic systems, and even locate specific U.S leaders should they so choose.”
Rep. Elissa Slotkin
In a speech on the House floor earlier this month, Slotkin noted that Chinese EVs, often sold much cheaper than their US and European counterparts, could quickly gain a significant share of the American market. She cited how Chinese vehicles, first sold in Europe in 2019, now make up almost a quarter of its market. The representative also recently pushed Secretary of the Army Christine Wormuth and Secretary of Defense Lloyd Austin on the security gap.
Alternatively (and perhaps ideally), legislators could pass a comprehensive data privacy law rather than dealing with these issues piecemeal.
The bill’s introduction follows the Biden Administration’s quadrupling of import tariffs on Chinese EVs. The White House’s new EV levies grew from 25 percent to 100 percent, following China’s EV exports rising 70 percent between 2022 and 2023.
In February, the White House also ordered the Department of Commerce to investigate the risks of connected vehicles from China and other adversaries. However, that action was conducted through an executive order and could be undone by future administrations. Slotkin’s legislation would close those loopholes if it makes it through Congress — rarely a safe bet in today’s highly obstructed and contentious political environment.
Russia has reportedly found new, more effective ways to knock out Ukraine’s Starlink service. The New York Timessaid on Friday that the increased interference has disrupted communications at critical moments and is posing “a major threat to Ukraine,” putting the country further on its heels more than two years into the war. How Russia is jamming Elon Musk’s satellite internet terminals is unclear.
The New York Times said Russia’s ability to jam communications has thrown off Ukraine’s ability to communicate, gather intelligence and conduct drone strikes. Ukrainian soldiers told the paper that jammed Starlink service stunts their ability to communicate quickly, leaving them scrambling to send text messages (often extremely slowly) to share intel about incoming or ongoing Russian maneuvers or attacks.
The jamming was reportedly repeated across Ukraine’s northern front line, often coinciding with Russian advances. The new outages are the first time Russia has jammed Starlink reception that widely and frequently. If it continues, it could “mark a tactical shift in the conflict,” highlighting Ukraine’s dependence on SpaceX’s internet technology. Without competing choices of similar quality, Volodymyr Zelenskyy’s democratic nation is left without many options that could work at the scale Ukraine needs.
Russia has tried to disrupt Ukraine’s comms since the war began, but Starlink service has reportedly held up well in the face of them. Something has changed. Ukraine’s digital minister, Mykhailo Federov, told The New York Times this week that Russia’s recent jamming appeared to use “new and more advanced technology.”
Federov told The NYT that Vladimir Putin’s army is now “testing different mechanisms to disrupt the quality of Starlink connections because it’s so important for us.” The digital minister didn’t specify the exact weapons Russia has been using, but a Russian official in charge of the country’s electronic warfare told state media last month that its military put Starlink on a “list of targets” and that it had developed ways to disrupt the service.
Ukraine President Volodymyr Zelenskyy (Armed Forces of Ukraine)
The disruptions highlight the power that one mercurial billionaire can have over the pivotal Eastern European war. Ukrainian officials have reportedly “appealed directly to Mr. Musk to turn on Starlink access during military operations” ahead of crucial drone strikes, and he hasn’t always obliged.
The Wall Street Journalreported in February that concern has grown that Musk could harbor at least some degree of Russian sympathies. He has posted comments on X that could be viewed as taking a pro-Russian stance, and disinformation experts worry that the way he runs the social platform could be friendly to Russian interference in the pivotal 2024 elections, including those in the US.
Musk spoke out earlier this year against the US sending more aid to Ukraine. Putin’s army also reportedly began using its own Starlink service, although Musk says he wasn’t aware of the terminals being sold to the Slavic nation. Ukrainian officials raised concerns earlier this year that Russia was buying Starlink tech from third-party vendors.
However, the Pentagon said earlier this month that the US has been “heavily involved in working with the government of Ukraine and SpaceX to counter Russian illicit use of Starlink terminals,” and a departing space official described SpaceX as “a very reliable partner” in those operations.
At I/O 2024, Google’s teaser for gave us a glimpse at where AI assistants are going in the future. It’s a multi-modal feature that combines the smarts of Gemini with the kind of image recognition abilities you get in Google Lens, as well as powerful natural language responses. However, while the promo video was slick, after getting to try it out in person, it’s clear there’s a long way to go before something like Astra lands on your phone. So here are three takeaways from our first experience with Google’s next-gen AI.
Sam’s take:
Currently, most people interact with digital assistants using their voice, so right away Astra’s multi-modality (i.e. using sight and sound in addition to text/speech) to communicate with an AI is relatively novel. In theory, it allows computer-based entities to work and behave more like a real assistant or agent – which was one of Google’s big buzzwords for the show – instead of something more robotic that simply responds to spoken commands.
Photo by Sam Rutherford/Engadget
In our demo, we had the option of asking Astra to tell a story based on some objects we placed in front of camera, after which it told us a lovely tale about a dinosaur and its trusty baguette trying to escape an ominous red light. It was fun and the tale was cute, and the AI worked about as well as you would expect. But at the same time, it was far from the seemingly all-knowing assistant we saw in Google’s teaser. And aside from maybe entertaining a child with an original bedtime story, it didn’t feel like Astra was doing as much with the info as you might want.
Then my colleague Karissa drew a bucolic scene on a touchscreen, at which point Astra correctly identified the flower and sun she painted. But the most engaging demo was when we circled back for a second go with Astra running on a Pixel 8 Pro. This allowed us to point its cameras at a collection of objects while it tracked and remembered each one’s location. It was even smart enough to recognize my clothing and where I had stashed my sunglasses even though these objects were not originally part of the demo.
In some ways, our experience highlighted the potential highs and lows of AI. Just the ability for a digital assistant to tell you where you might have left your keys or how many apples were in your fruit bowl before you left for the grocery store could help you save some real time. But after talking to some of the researchers behind Astra, there are still a lot of hurdles to overcome.
Photo by Sam Rutherford/Engadget
Unlike a lot of Google’s recent AI features, Astra (which is described by Google as a “research preview”) still needs help from the cloud instead of being able to run on-device. And while it does support some level of object permanence, those “memories” only last for a single session, which currently only spans a few minutes. And even if Astra could remember things for longer, there are things like storage and latency to consider, because for every object Astra recalls, you risk slowing down the AI, resulting in a more stilted experience. So while it’s clear Astra has a lot of potential, my excitement was weighed down with the knowledge that it will be some time before we can get more full-feature functionality.
Karissa’s take:
Of all the generative AI advancements, multimodal AI has been the one I’m most intrigued by. As powerful as the latest models are, I have a hard time getting excited for iterative updates to text-based chatbots. But the idea of AI that can recognize and respond to queries about your surroundings in real-time feels like something out of a sci-fi movie. It also gives a much clearer sense of how the latest wave of AI advancements will find their way into new devices like smart glasses.
Google offered a hint of that with Project Astra, which may one day have a glasses component, but for now is mostly experimental (the video during the I/O keynote were apparently a “research prototype.”) In person, though, Project Astra didn’t exactly feel like something out of sci-fi flick.
Photo by Sam Rutherford/Engadget
It was able to accurately recognize objects that had been placed around the room and respond to nuanced questions about them, like “which of these toys should a 2-year-old play with.” It could recognize what was in my doodle and make up stories about different toys we showed it.
But most of Astra’s capabilities seemed on-par with what Meta has available with its smart glasses. Meta’s multimodal AI can also recognize your surroundings and do a bit of creative writing on your behalf. And while Meta also bills the features as experimental, they are at least broadly available.
The Astra feature that may set Google’s approach apart is the fact that it has a built-in “memory.” After scanning a bunch of objects, it could still “remember” where specific items were placed. For now, it seems Astra’s memory is limited to a relatively short window of time, but members of the research team told us that it could theoretically be expanded. That would obviously open up even more possibilities for the tech, making Astra seem more like an actual assistant. I don’t need to know where I left my glasses 30 seconds ago, but if you could remember where I left them last night, that would actually feel like sci-fi come to life.
But, like so much of generative AI, the most exciting possibilities are the ones that haven’t quite happened yet. Astra might get there eventually, but right now it feels like Google still has a lot of work to do to get there.
Catch up on all the news from Google I/O 2024 right here!
AlphaFold software, from Google DeepMind and (the also Alphabet-owned) Isomorphic Labs, has already demonstrated that it can predict how proteins fold with shocking accuracy. It’s cataloged a staggering 200 million known proteins, and Google says millions of researchers have used previous versions to make discoveries in areas like malaria vaccines, cancer treatment and enzyme designs.
Knowing a protein’s shape and structure determines how it interacts with the human body, allowing scientists to create new drugs or improve existing ones. But the new version, AlphaFold 3, can model other crucial molecules, including DNA. It can also chart interactions between drugs and diseases, which could open exciting new doors for researchers. And Google says it does so with 50 percent better accuracy than existing models.
“AlphaFold 3 takes us beyond proteins to a broad spectrum of biomolecules,” Google’s DeepMind research team wrote in a blog post. “This leap could unlock more transformative science, from developing biorenewable materials and more resilient crops, to accelerating drug design and genomics research.”
“How do proteins respond to DNA damage; how do they find, repair it?” Google DeepMind project leader John Jumper toldWired. “We can start to answer these questions.”
Before AI, scientists could only study protein structures through electron microscopes and elaborate methods like X-ray crystallography. Machine learning streamlines much of that process by using patterns recognized from its training (often imperceptible to humans and our standard instruments) to predict protein shapes based on their amino acids.
Google says part of AlphaFold 3’s advancements come from applying diffusion models to its molecular predictions. Diffusion models are central pieces of AI image generators like Midjourney, Google’s Gemini and OpenAI’s DALL-E 3. Incorporating these algorithms into AlphaFold “sharpens the molecular structures the software generates,” as Wiredexplains. In other words, it takes a formation that looks fuzzy or vague and makes highly educated guesses based on patterns from its training data to clear it up.
“This is a big advance for us,” Google DeepMind CEO Demis Hassabis told Wired. “This is exactly what you need for drug discovery: You need to see how a small molecule is going to bind to a drug, how strongly, and also what else it might bind to.”
AlphaFold 3 uses a color-coded scale to label its confidence level in its prediction, allowing researchers to exercise appropriate caution with results that are less likely to be accurate. Blue means high confidence; red means it’s less certain.
Google is making AlphaFold 3 free for researchers to use for non-commercial research. However, unlike with past versions, the company isn’t open-sourcing the project. One prominent researcher who makes similar software, University of Washington professor David Baker, expressed disappointment to Wired that Google chose that route. However, he was also wowed by the software’s capabilities. “The structure prediction performance of AlphaFold 3 is very impressive,” he said.
As for what’s next, Google says “Isomorphic Labs is already collaborating with pharmaceutical companies to apply it to real-world drug design challenges and, ultimately, develop new life-changing treatments for patients.”
Microsoft Research Asia has unveiled a new experimental AI tool called VASA-1 that can take a still image of a person — or the drawing of one — and an existing audio file to create a lifelike talking face out of them in real time. It has the ability to generate facial expressions and head motions for an existing still image and the appropriate lip movements to match a speech or a song. The researchers uploaded a ton of examples on the project page, and the results look good enough that they could fool people into thinking that they’re real.
While the lip and head motions in the examples could still look a bit robotic and out of sync upon closer inspection, it’s still clear that the technology could be misused to easily and quickly create deepfake videos of real people. The researchers themselves are aware of that potential and have decided not to release “an online demo, API, product, additional implementation details, or any related offerings” until they’re sure that their technology “will be used responsibly and in accordance with proper regulations.” They didn’t, however, say whether they’re planning to implement certain safeguards to prevent bad actors from using them for nefarious purposes, such as to create deepfake porn or misinformation campaigns.
The researchers believe their technology has a ton of benefits despite its potential for misuse. They said it can be used to enhance educational equity, as well as to improve accessibility for those with communication challenges, perhaps by giving them access to an avatar that can communicate for them. It can also provide companionship and therapeutic support for those who need it, they said, insinuating the VASA-1 could be used in programs that offer access to AI characters people can talk to.
According to the paper published with the announcement, VASA-1 was trained on the VoxCeleb2 Dataset, which contains “over 1 million utterances for 6,112 celebrities” that were extracted from YouTube videos. Even though the tool was trained on real faces, it also works on artistic photos like the Mona Lisa, which the researchers amusingly combined with an audio file of Anne Hathaway’s viral rendition of Lil Wayne’s Paparazzi. It’s so delightful, it’s worth a watch, even if you’re doubting what good a technology like this can do.
This article contains affiliate links; if you click such a link and make a purchase, we may earn a commission.
In addition to the supermassive black hole at the center of our galaxy, the Milky Way also serves as home to smaller stellar black holes that form when a massive star collapses. Scientists believe there are 100 million stellar black holes in our galaxy alone, but most of them have yet to be discovered. The ones that had already been found are, on average, around 10 times the size of our sun, with the biggest one reaching 21 solar masses. Thanks to the information collected by the European Space Agency’s Gaia mission, though, scientists have discovered a stellar black hole that’s 33 times the size of our sun, making it the biggest one of its kind we’ve ever seen in our galaxy so far. It’s also relatively close to our planet at around 1,926 light-years away.
Gaia BH3, as it’s now called, was first noticed by a team of ESA scientists poring over data from the mission to look for anything unusual. An old giant star from the nearby Aquila constellation caught their attention with its wobbling, leading to the discovery that it was orbiting a massive black hole. BH3 was hard to find despite being so close — it’s now the second closest known black hole to our planet — because it doesn’t have celestial bodies close enough that could feed it matter and make it light up in X-ray telescopes. Before its discovery, we’d only found black holes of comparable size in distant galaxies.
The ESA team used data from ground-based telescopes like the European Southern Observatory to confirm the size of the newly discovered celestial body. They also published a paper with preliminary findings before they release a more detailed one in 2025, so that their peers could start studying Gaia BH3. For now, what they know is that the star orbiting it has very few elements heavier than hydrogen and helium, and since stellar pairs tend to have similar compositions, the star that collapsed to form BH3 could’ve been the same.
Scientists have long believed that it’s the metal-poor stars that can create high-mass black holes after they collapse, because they lose less mass in their lifetimes. In other words, they’d theoretically still have a lot of materials left by the time of their death to form a massive black hole. This was apparently the first evidence we’ve found that links metal-poor stars with massive stellar black holes, and it’s also proof that older giant stars developed differently than the newer ones we see in our galaxy.
We’ll most likely see more detailed studies about binary systems and stellar black holes that use data from BH3 and its companion star in the future. The ESA believes that BH3’s discovery is just the beginning, and it’s going to be the focus of more investigations as we seek to unravel the mysteries of the universe.
Three companies are vying for the opportunity to send their own lunar vehicle to the moon to support NASA’s upcoming Artemis missions. The this week that it’s chosen Intuitive Machines, Lunar Outpost and Venturi Astrolab to develop their lunar terrain vehicles (LTV) in a feasibility study over the next year. After that, only one is expected to be selected for a demonstration mission, in which the vehicle will be completed and sent to the moon for performance and safety tests. NASA is planning to use the LTV starting with the Artemis V crew that’s projected to launch in early 2030.
The LTV that eventually heads to the moon’s south pole needs to function as both a crewed and uncrewed vehicle, serving sometimes as a mode of transportation for astronauts and other times as a remotely operated explorer. NASA says it’ll contract the chosen vehicle for lunar services through 2039, with all the task orders relating to the LTV amounting to a potential value of up to $4.6 billion. The selected company will also be able to use its LTV for commercial activities in its down time.
Lunar Outpost
Astrolab
Intuitive Machines, which will be developing an LTV called the Moon Racer, has already bagged multiple contracts with NASA as part of the Commercial Lunar Payload Services (CLPS) program, and , Odysseus, to the moon to achieve . Venturi Astrolab will be developing a vehicle it’s dubbed Flex, while Lunar Outpost will be working on an LTV called Lunar Dawn. All must be able to support a crew of two astronauts and withstand the extreme conditions of the lunar south pole.
“We will use the LTV to travel to locations we might not otherwise be able to reach on foot, increasing our ability to explore and make new scientific discoveries,” said Jacob Bleacher, a chief exploration scientist at NASA.
Weather Update, April 7, 4:00 AM ET: The weather forecast in the story below still largely holds, but things are more unsettled in the southern US, with forecasts now calling for thunderstorms from Dallas up to Indianapolis. So, keep that potential danger in mind (and keep an eye on forecasts) when making eclipse plans.
Elsewhere, the best chance of good viewing along the path of totality is still in northeastern parts of the US (Buffalo, NY, Burlington, VT), along with southeast Canada (Niagara Falls and Montreal), according to Accuweather. In the Midwest (Cleveland), there’s a higher chance of rain than before (58 percent), but no storms currently predicted.
Original story continues below
On April 8, a solar eclipse will darken the skies. This is a rare astronomical event: The last North American total solar eclipse was on August 21, 2017, and there won’t be another on visible on the continent until 2044. The path of totality — where the sun will be fully blocked by the moon — covers over 30 million people in the US, Canada and Mexico.
Those lucky folks may see the sun’s corona and a “diamond ring” — both dramatic sights. Other regions will experience partial eclipses, with the level depending on how close you are to totality. Watching the moon eat into the sun, even a bit, is still a spectacular sight.
So, which cities and regions will experience totality, and when? What’s the weather forecast in those areas? And if you do have a clear view, how can you safely watch and record the event?
Where in the US will you experience the solar eclipse totality, and when?
The good news is that many major centers are in the 100-mile-wide band of totality, so millions of people will be able to see a full solar eclipse. It follows a northeast path, so Mexico’s Pacific coast will get the first views in Mazatlan starting at around 10:57 AM PDT (total eclipse starting at 12:07 PDT), followed by the city of Torreón (all times local).
The total eclipse moves into the United states at 12:10 PM CDT (Eagle Pass, Texas), then hits Austin, Fort Worth and Dallas — three out of five of the most populous Texas cities. From there, it moves into Little Rock, Arkansas, followed by select parts of Missouri, Illinois and Indiana (including Indianapolis).
NASA
Ohio cities Dayton, Toledo and Cleveland get the full show, followed by Erie, Pennsylvania, then Buffalo, Rochester and Syracuse in New York along with Maine. Canada is in on the fun too, with parts of southern Ontario (Hamilton, Niagara Falls) and Quebec (Montreal) getting the totality, along with New Brunswick, PEI and finally, Bonavista, Labrador at 4:03 PM NDT (Newfoundland Time).
If you’re elsewhere on the continent and can’t travel, know that the closer you are to the band of totality, the more the sun will be obscured by the moon (this map shows how much of the eclipse you’ll get depending where you are on the continent).
An impressive list of major centers are within 200 miles of totality, so they’ll get a 90 percent or better eclipse (Houston, St. Louis, Memphis, Nashville, Chicago, Cincinnati, Detroit, Toronto, New York, Boston).
Anyone in the US south, midwest and northeast should get a decent spectacle, as will folks in Canada’s southeast and Atlantic coast. Even if you’re not in those regions, you might still see (and can capture) a mini eclipse.
How long with the 2024 solar eclipse last?
From the beginning when the moon first starts to cover the sun (partial eclipse) until the end when the two bodies part ways is a good long time – up to two hours and forty minutes in Dallas, and 2:18 in Caribou, Maine.
However, totality itself is brief, with the duration dependent on how close you are to the center of the totality band and the time of day. It’s at just under four minutes in Dallas, less than three minutes in Presque Island, Maine and a mere minute and 12 seconds in Montreal. As such, you’ll need to be ready and hope that the skies are clear during that brief window.
What’s the weather forecast in my area?
It’s still early for an accurate forecast, but a week is enough to get a general idea by region. Suffice to say, April isn’t the ideal month for clear skies. That said, an eclipse can still be visible through light cloud cover, and even if it’s thick, the sky will grow dramatically dark.
Unfortunately, the odds of precipitation are indeed above average across most of the band of the eclipse. Forecasts predict that the chances for clear skies are better the farther northeast you live, the opposite of historical trends.
To wit, Dallas has showers forecast throughout the day (58 percent), which would mean continuous cloud cover and no clear view if that holds. That improves a bit when you get to Indianapolis (partly cloudy, 24 percent chance of rain), with things better still in Buffalo, New York (partly cloudy, 11 percent).
Things are looking good right now in Montreal, though, with mostly sunny skies and only a 9 percent chance of rain, and the same goes for Fredericton, New Brunswick.
Niagara Falls also figures to have decent weather during the eclipse (mostly sunny, 18 percent) and is in the path of totality, which has led to the city declaring a state of emergency out of caution. Officials estimate that a million people could pour into the area, creating potentially dangerous crowds.
How can I watch the solar eclipse at home?
Staring at the sun is obviously dangerous for your vision, and doing so during an eclipse can be just as harmful. Even though you may not feel discomfort immediately, you may damage your eyes via an affliction called solar retinopathy. That can lead to serious consequences like eye pain, blind spots, blurred vision and more.
American Astronomical Society
To view it safely, you must purchase a pair of approved solar eclipse glasses based on an international safety standard called ISO 12312-2 (regular sunglasses won’t do). That dictates the maximum luminous transmittance, along with the range of permissible wavelength transmittance (UVA, UVB and infrared).
There’s certainly still time to grab a pair if you don’t have them already. The American Astronomical Society (AAS) has many recommendations for manufacturers and vendors, both online and at retail chains.
Warby Parker, for one, is offering free glasses (limit two per person while supplies last). You can also find them at Staples, Lowes and Walmart, or online at B&H and multiple science and astronomy stores.
The AAS advises against searching for the lowest price on Amazon or eBay, however, in case you get a bad knock off. “Before you buy a solar viewer or filter online, we recommend that you make sure that (1) the seller is identified on the site and (2) the seller is listed on this page,” it says on its Solar Eclipse Across America site.
How to watch the solar eclipse safely without glasses
Canadian Space Agency
It’s possible to view an eclipse without glasses via indirect means, as well. The simplest way is by punching a small round hole in a piece of thick paper or cardboard, then positioning it so the sun shines through the hole onto the ground or a flat surface (you can also attach a piece of foil with a hole, as NASA shows here). That will project an image of the Sun’s disc, letting you see the eclipse in real time.
The same pinhole principle would let you use anything with perforated holes, like a colander, projecting dozens of tiny eclipses on a surface. Trees can do the same thing, casting weird leaf shadows with little solar eclipse chunks out of them.
Benjamin Seigh/Wikimedia
For a bit better experience, you can build a crude box projector. With that, the sun shines through a hole in tin foil onto a white card, and you can look through a larger hole at the card, with the sun behind you. The Canadian Space Agency explains exactly how to make that.
Never, ever view an eclipse directly through a pair of binoculars or a telescope, as that’s a guaranteed way to damage your eyes. That said, you can use a pair of binoculars or a telescope to project the sun onto a piece of paper, as shown in this video.
How to take photos or video of the solar eclipse
Unfortunately, you can’t just point your smartphone or camera at the sun to record the eclipse, as the brightness will overwhelm the sensor and ruin the image (and possibly damage the sensor). Luckily, you can shield your camera just as you do your eyes.
The cheapest way to do that is to buy an extra set of eclipse glasses, then cut out an eyepiece from one and tape it over the smartphone (or other camera) lens. That will reduce the light levels enough to see detail in the sun throughout the partial eclipse and totality.
You can also purchase dedicated smartphone solar filters like the VisiSolar Photo Filter, which are designed for cameras and not direct viewing. Another choice is the Solar Snap Eclipse App Kit, which also offers an app that aids in photographing the eclipsed sun. It’s advisable to also wear solar glasses when setting up your smartphone or camera to protect your eyes.
If you’re shooting the eclipse with a dedicated mirrorless or DSLR camera, you’ll need either a mylar, 16-stop ND (neutral density) or hydrogen alpha solar filter. Again, do not look directly into a DSLR’s optical viewfinder at the sun if the lens doesn’t have one of those filters attached (the electronic viewfinder on a mirrorless camera is safe).
To photograph the eclipse with a smartphone, turn the flash off and put the camera into ultrawide or wide mode so it stays in frame. Do NOT look directly at the sun to line up your camera if you’re not wearing solar eclipse glasses.
Don’t use the digital zoom to try to make the eclipse bigger, as you’ll lose resolution (you can zoom in later in your photo editing app). Once focus is set on the sun, use your smartphone’s focus lock feature so that it doesn’t “hunt” for focus and blur the eclipse.
During totality, the “diamond ring” effect only lasts a split second, so use the burst mode of your camera or you’ll likely miss the shot. And try to capture RAW (rather than JPEG) images to keep the maximum detail possible for later editing. Some iPhone and Android smartphones have RAW capability built-in, if not, you can use a third-party app.
If you decide to capture video, you’ll need a filter as well, of course. But you should also use a tripod, as shooting handheld will induce blur and result in a shaky video. Even a cheap tripod will do the trick, along with a simple smartphone holder. Capture the highest resolution you can (4K or even 8K) at the highest quality possible. You’ll also capture any cheering, shouting, etc. — a precious souvenir you can look back on again and again.
More resources
There are plenty of government and private sites with more information about the eclipse, starting with the AAS’s eclipse site, detailing things like eye safety, imaging, resources and even a totality app — an “interactive map that shows what you’ll see at any location in North America for the total solar eclipse of April 8, 2024.”
Tesla is introducing a robotaxi on August 8, Elon Musk has announced on X a few hours after Reuters published a report that the automaker is scrapping its plans to produce a low-cost EV. Reuters also said that Musk’s directive was to “go all in” on robotaxis built on the company’s small-vehicle platform. Tesla has been promising a more affordable EV with prices expected to start at $25,000 for years, and Musk said as recently as this January that he’s optimistic the model will arrive in the second half of 2025. In response to the report, the Tesla chief tweeted that “Reuters is lying (again).”
He didn’t clarify which part of the report was a lie, but considering he confirmed that Tesla is unveiling a robotaxi, he likely meant the news organization’s claim that the company pulled the plug on a more affordable EV. At the moment, Tesla’s cheapest vehicle is the Model 3, but its prices start at $39,000. It’ll be interesting to see how the company will make a robotaxi work with its camera-only system — it dropped radar and other sensors, which robotaxi companies like Waymo use extensively, from its driver assistance technologies a few years ago.
Gig work predates the internet. Besides traditional forms of self-employment, like plumbing, offers for ad-hoc services have long been found in the Yellow Pages and newspaper classified ads, and later Craigslist and Backpage which supplanted them. Low-cost broadband internet allowed for the proliferation of computer-based gig platforms like Mechanical Turk, Fiverr and Elance, which offered just about anyone some extra pocket change. But once smartphones took off, everywhere could be an office, and everything could be a gig — and thus the gig economy was born.
Maybe it was a confluence of technological advancement and broad financial anxiety from the 2008 recession, but prospects were bad, people needed money and many had no freedom to be picky about how. This was the same era in which the phrase “the sharing economy” proliferated — at once sold as an antidote to overconsumption, but that freedom from ownership belied the more worrying commoditization of any skill or asset. Of all the companies to take advantage of this climate, none went further or have held on harder than Uber.
Uber became infamous for railroading its way into new markets without getting approval from regulators. It cemented its reputation as a corporate ne’er-do-well through a byzantine scandal to avoid regulatory scrutiny, several smaller ones over user privacy and minimally-beneficial surcharges as well as, in its infancy, an internal reputation for sexual harassment and discrimination. Early on, the company used its deep reserves of venture capital to subsidize its own rides, eating away at the traditional cab industry in a given market, only to eventually increase prices and try to minimize driver pay once it reached a dominant position. Those same reserves were spent aggressively recruiting drivers with signup bonuses and convincing them they could be their own boss.
Self-employment has a whiff of something liberatory, but Uber effectively turned a traditionally employee-based industry into one that was contractor-based. This meant that one of the first casualties of the ride-sharing boom were taxi medallions. For decades, cab drivers in many locales effectively saw these licenses as retirement plans, as they’d be able to sell them on to newcomers when it was time to hang up their flat cap. But in large part due to the influx of ride-sharing services, the value of medallions has plummeted over the last decade or so — in New York, for instance, the value of a medallion dropped from around $1 million in 2014 to $100,000 in 2021. That’s in tandem with a drop in earnings, leaving many struggling to pay off enormous loans they took out to buy a medallion.
Some jurisdictions have sought to offset that collapse in medallion value. Quebec pledged $250 million CAD in 2018 to compensate cab drivers. Other regulators, particularly in Australia, applied a per-ride fee to ride-sharing services as part of efforts to replace taxi licenses and compensate medallion holders. In each of those cases, taxpayers and riders, not rideshare companies, bore the brunt of the impact on medallion holders.
At first it was just cab drivers that were hurting, but over the years, compensation for this new class of non-employee app drivers dried up too. In 2017, Uber paid $20 million to settle allegations from the Federal Trade Commission that it used false promises about potential earnings to entice drivers to join its platform. Late last year, Uber and Lyft agreed to pay $328 million to New York drivers after the state conducted a wage theft investigation. The settlement also guaranteed a minimum hourly rate for drivers outside of New York City, where drivers were already subject to minimum rates under Taxi & Limousine Commission rules.
Many rideshare drivers have also sought recognition as employees rather than contractors, so they can have a consistent hourly wage, overtime pay and benefits — efforts that the likes of Uber and rival Lyft have been fighting against. In January, the Department of Labor issued a final rule that aims to make it more difficult for gig economy companies to classify workers as independent contractors rather than employees. The EU is also weighing a provisional deal to reclassify millions of app workers as employees.
Of course, the partial erosion of an entire industry’s labor market wasn’t always the end goal. At one point, Uber wanted to zero out labor costs by getting rid of drivers entirely. It planned to do so by rolling out a fleet of self-driving vehicles and flying taxis.
“The reason Uber could be expensive is because you’re not just paying for the car — you’re paying for the other dude in the car,” former CEO Travis Kalanick said in 2014, a day after Uber suggested drivers could make $90,000 per year on the platform. “When there’s no other dude in the car, the cost of taking an Uber anywhere becomes cheaper than owning a vehicle. So the magic there is, you basically bring the cost below the cost of ownership for everybody, and then car ownership goes away.”
Uber’s grand automation plans didn’t work out as intended, however. The company, under current CEO Dara Khosrowshahi, sold its self-driving car and flying taxi units in late 2020.
Uber’s success had second-order effects too: despite a business model best described as “set money on fire until (fingers crossed!) a monopoly is established” a whole slew of startups were born, taking their cues from Uber or explicitly pitching themselves as “Uber for X.” Sure, you might find a place to stay on Airbnb or Vrbo that’s nicer and less expensive than a hotel room. But studies have shown that such companies have harmed the affordability and availability of housing in some markets, as many landlords and real-estate developers opt for more profitable short-term rentals instead of offering units for long-term rentals or sale. Airbnb has faced plenty of other issues over the years, from a string of lawsuits to a mass shooting at a rental home.
Increasingly, this is becoming the blueprint. Goods and services are exchanged by third parties, facilitated by a semi-automated platform rather than a human being. The platform’s algorithm creates the thinnest veneer between choice and control for the workers who perform identical labor to the industry that platform came to replace, but that veneer allows the platform to avoid traditionally pesky things like legal liability and labor laws. Meanwhile, customers with fewer alternative options find themselves held captive by these once-cheap platforms that are now coming to collect their dues. Dazzled by the promise of innovation, regulators rolled over or signed a deal with the devil. It’s everyone else who’s paying the cost.
To celebrate Engadget’s 20th anniversary, we’re taking a look back at the products and services that have changed the industry since March 2, 2004.
A budding startup called Interlune is trying to become the first private company to mine the moon’s natural resources and sell them back on Earth. Interlune will initially focus on helium-3 — a helium isotope created by the sun through the process of fusion — which is abundant on the moon. In an interview with Ars Technica, Rob Meyerson, one of Interlune’s founders and former Blue Origin president, said the company hopes to fly its harvester with one of the upcoming commercial moon missions backed by NASA. The plan is to have a pilot plant on the moon by 2028 and begin operations by 2030, Meyerson said.
Interlune announced this week that it’s raised $18 million in funding, including $15 million in its most recent round led by Seven Seven Six, the venture firm started by Reddit co-founder Alexis Ohanian. The resource it’s targeting, helium-3, could be used on Earth for applications like quantum computing, medical imaging and, perhaps some day down the line, as fuel for fusion reactors. Helium-3 is carried to the moon by solar winds and is thought to remain on the surface trapped in the soil, whereas when it reaches Earth, it’s blocked by the magnetosphere.
Interlune aims to excavate huge amounts of the lunar soil (or regolith), process it and extract the helium-3 gas, which it would then ship back to Earth. Alongside its proprietary lunar harvester, Interlune is planning a robotic lander mission to assess the concentration of helium-3 at the selected location on the surface.
Interlune
“For the first time in history,” Meyerson said in a statement, “harvesting natural resources from the Moon is technologically and economically feasible.” The founding team includes Meyerson and former Blue Origin Chief Architect Gary Lai, Apollo 17 astronaut Harrison H. Schmitt, former Rocket Lab exec Indra Hornsby and James Antifaev, who worked for Alphabet’s high-altitude balloon project, Loon.
Mini has promised to , a date that seemed pretty far off back in 2021 but right now is starting to sound not that far off at all. While the company’s prior battery-powered efforts have been great, it’s going to take something more serious and more practical to convert the masses to the wonders of electrification.
That something might just be the 2025 Mini Countryman. While Mini will offer this car with a gasoline-burning engine if you’re feeling traditional, the star of the lineup will be the new, $45,200, all-electric Countryman SE. With 313 horsepower and 363 pound-feet of torque, it’s quick, and with way more cargo space, it’s practical too.
But how does it drive? That’s what we headed to Portugal to find out, and while the extra volume and weight of the new Countryman does come with some compromises, it’s an engaging SUV to drive with a fantastic interior that’s just a few software updates away from perfection. Full the full preview, watch the video up top.
