Artificial intelligence-powered video maker Runway has added the promised image-to-video feature to its Gen-3 model released a few weeks ago, and it may be as impressive as promised. Runway has enhanced the feature to address its biggest limitations in the Gen-2 model released early last year. The upgraded tool is miles better at character consistency and hyperrealism, making it a more powerful tool for creators looking to produce high-quality video content.
Runway's Gen-3 model is still in alpha testing and only available to subscribers who pay $12 per month per editor for the most basic package. The new model had already attracted plenty of interest even when it came out with only text-to-video capabilities. But, no matter how good a text-to-video engine is, it has inherent limits, especially when it comes to characters in a video looking the same over multiple prompts and appearing to be in the real world. Without visual continuity, it's hard to make any kind of narrative. In earlier iterations of Runway, users often struggled to keep characters and settings uniform across different scenes when relying solely on text prompts.
Offering reliable consistency in character and environmental design is no small thing, but the use of an initial image as a reference point to maintain coherence across different shots can help. In Gen-3, Runway's AI can create a 10-second video guided by additional motion or text prompts in the platform. You can see how it works in the video below.
Stills to Films
Runway's image-to-video feature doesn't just ensure people and backgrounds stay the same when seen from a distance. Gen-3 also incorporates Runway's lip-sync feature so that someone speaking moves their mouth in a way that matches the words they are saying. A user can tell the AI model what they want their character to say, and the movement will be animated to match. Combining synchronized dialogue and realistic character movements will interest a lot of marketing and advertising developers looking for new and, ideally, cheaper ways to produce videos.
Runway isn't done adding to the Gen-3 platform, either. The next step is bringing the same enhancements to the video-to-video option. The idea is to keep the same motion but in a different style. A human running down a street becomes an animated anthropomorphic fox dashing through a forest, for instance. Runway will also bring its control features to Gen-3, such as Motion Brush, Advanced Camera Controls, and Director Mode.
AI video tools are still in the early stages of development, with most models excelling in short-form content creation but struggling with longer narratives. That puts Runway and its new features in a strong position from a market standpoint, but it is far from alone. Midjourney, Ideogram, Leonardo (now owned by Canva), and others are all racing to make the definitive AI video generator. Of course, they're all keeping a wary watch on OpenAI and its Sora video generator. OpenAI has some advantages in name recognition, among other benefits. In fact, Toys"R"Us has already made a short film commercial using Sora and premiered it at the Cannes Lions Festival. Still, the film about AI video generators is only in its first act, and the triumphant winner cheering in slow-motion at the end is far from inevitable.
Cybercriminals found a way to send millions of “perfectly spoofed” phishing emails thanks to a vulnerability in Proofpoint’s email relay servers.
Experts from Guardio Labs revealed the phishing campaign started in January 2024, and was sending out an average of three million emails daily. In early June, it peaked with 14 million emails being disseminated.
The researchers dubbed the campaign “EchoSpoofing”, noting the crooks were able to get their phishing emails properly DKIM signed, and SPF approved. What tipped researchers off, though, was that all of the emails were dispatched from one specific family of relay servers - pphosted.com - which is owned and operated by the email security vendor Proofpoint.
Bypassing spam filters
To the recipient, the email looks as if it is coming from a legitimate business. The businesses being spoofed here all seem to be Proofpoint’s customers, mostly Fortune 100 companies. These include Disney, IBM, Nike, Best Buy, and Coca-Cola, to name a few.
“These emails echoed from official Proofpoint email relays with authenticated SPF and DKIM signatures, thus bypassing major security protections — all to deceive recipients and steal funds and credit card details,” the researchers concluded.
Guardio Labs said that all major email platforms, including Gmail, failed to flag these emails as spam, and instead allowed them straight into people’s inboxes. The emails scared the victims with fake account expirations, payment and renewal requests, and similar, all with the goal of harvesting payment and personally identifiable information.
Proofpoint said it had been keeping an eye on the EchoSpoofing campaign since March 2024, and that it provided new settings, and advice, on how to prevent such attacks in the future. The company provided a detailed guide on how users can add anti-spoof checks, and more.
As you're getting ready to hit the beach or travel to the Paris Olympics, beware: travel applications are among the most data-hungry and privacy-unfriendly apps.
Cybersecurity experts at NordVPN, one of the best VPN providers on the market, have found the likes of Booking, Airbnb, and FlightRadar24 request some of the most unnecessary permissions to access your device functions. Put simply, they want to access your data even when it's unrelated to fulfilling their performance, including some of the most sensitive information like your biometrics details.
All this is bad for your privacy, especially when considering the ever-growing threat of data breaches involving big companies. Booking.com, for instance, was the target of a major phishing campaign last year. While it may be unimaginable to plan your holiday without one of these apps, there are still some actions you can take to protect your privacy when using them.
Travel apps: "Over 28% of permissions are excessive"
"Travel apps are among the most eager to request access to devices' functions that are not needed for their performance. Collected data could be used against the user's interests and lead to privacy issues that are way more serious than targeted ads," said Adrianus Warmenhoven, cybersecurity advisor at NordVPN.
According to researchers, every travel app asks for an average of 23 device permissions, including access to your travel location, photos, or videos. More than six of these permissions are unnecessary, though, as the application doesn't need such details for its functionality.
The travel category was also found to be particularly eager to access your biometrics details. This is a highly sensitive piece of information as it deals with system-critical processes - think of device unblocking, multi-factor authentication, or online payments. On average, travel apps request nine special, dangerous, and biometric permissions.
Besides travel apps, social networking, health and lifestyle, and navigation applications are some of the most unfriendly apps for data access and unnecessary permission requests. You can read the full findings of NordVPN's research here.
How to stay safe when using travel apps
Let's face it, in today's digitalized world it would be foolish to think that we can avoid using some travel apps when we plan or go on holiday. Yet, we can still do so while protecting our privacy.
Did you know?
A virtual private network (VPN) is also a very handy tool when traveling. Besides protecting your data and devices when connecting to unsecured public Wi-Fi, a travel VPN also allows you to keep streaming all your favorite shows while traveling abroad.
As a rule of thumb, Warmenhoven recommends downloading travel apps (and any other application, for that matter) only from official stores or websites. This is because, he explains, unofficial app stores won't always have systems to check whether an app is safe before it's published and available to download. Criminals are also infamous for using insecure applications to spread malware attacks.
According to Warmenhoven, you should also take some time to familiarise yourself with the app's privacy policy before downloading it. He especially recommends checking what information the provider will record about you and share with third parties.
As we have seen, reviewing your app's data permissions is also crucial. "Users should always consider whether the app needs certain data to do its job before tapping 'Accept,' even if the app is developed by a well-known and trustworthy traveling service provider," said Warmenhoven, suggesting particular attention to permissions like camera, microphone, storage, location, and contact list.
You should avoid automatically signing in with social network accounts, too. If you do so, the app will be able to collect information from the account and vice versa.
Lastly, remember to delete apps you don't use. Why risk your privacy for an application you're not even using?
from TechRadar - All the latest technology news https://ift.tt/C9ijf06
Artificial intelligence can produce impressive images, but it isn't uncommon for these images to have weird problems, such as people with too many teeth or cityscapes with Escher-style street layouts. Google Gemini is working on upgrading its AI image creation feature to fix those sorts of problems, as first spotted in unfinished code by Android Authority. It appears a fine-tuning capability is on its way, which will allow users to make detailed edits to their AI-generated images.
Google Gemini's text-to-image tools can't make edits after creating the image right now. Instead, users have to submit new prompts, hoping the new prompt will fix any problems and create something that matches what they want to see. That can be especially tedious if there's only a small but still distracting error. According to the uncovered code, Gemini's fine-tuning feature will address the need for limited changes with two editing methods.
The first option will let users submit a prompt about an AI-generated image and ask for a change to one aspect. For instance, if you liked the image above but wanted to set it in a city, you could keep the robot and bird but change the background by asking Gemini to move them. The second method described in the code is a more interactive approach. Users could circle the part of the image they want to change using their finger or a stylus. Once the area is selected, they can describe the desired changes, and Gemini will understand that the instructions pertain only to the circled section.
AI Editing Success
These editing tools could particularly benefit those in fields such as graphic design, marketing, and social media, where visual accuracy and quick turnaround times are crucial. Google Gemini can better serve the needs of artists, designers, and casual users who seek to create polished visual content more efficiently. While the exact release date of these features remains uncertain, their appearance in the code suggests it won't be long coming. It also pairs well with related features like the upcoming Ask Photos image search feature.
Google won't be the first to deploy editing tools to AI image makers. These methods are largely the same as those available with OpenAI's Dall-E portfolio of AI image-making models. In ChatGPT, users can ask for adjustments to an already produced image, or they can highlight parts of it and submit a new text prompt adjusting that part of the picture. There are similar features for many AI image creators like Ideogram.ai and Adobe Firefly. Still, Google's plan to incorporate these fine-tuning tools is a technical jump for Gemini. It marks Google's ongoing push to match and surpass its rivals at OpenAI, Meta, and elsewhere when it comes to generative AI tools.
Recently I attended an HP Imagine AI event, which covered the full scope of the tech giant’s plans for AI in both its software and hardware. This was one of many recent events, showcases, and hands-on from numerous manufacturers who are clamoring to show media and buyers what it can offer in this increasingly overcrowded market.
While HP’s offerings were quite interesting and often seemed quite useful, this all poses a much larger question: what is the purpose of all this? A second question must be begged afterward: what is the true future for AI? Once you unravel all the glam of these AI events, the shiny new AI PCs, and the multitude of AI-powered software and tools, this is the foundational question left.
After ruminating on it for a while, I believe I found my deceptively simplistic answer: accessibility—a simple answer, but a truly profound one that draws out AI’s true potential.
(Image credit: Future)
What did HP reveal?
HP first delved into its mission statement regarding AI — to make AI real it must be tangible, show HP’s leadership, and catalyze the industry. The manufacturer also made it clear that one of its main goals would be to address audience concerns regarding trust and establishing a verification system in AI models. It also needs to address issues such as hallucination, drift, and bias in its AI data sets. Whether or not HP can live up to these ideals will be left to the test of time, however.
One of the main reveals was the HP OmniBook Ultra 14, which offers an AMD Ryzen AI 300 SoC with 55 TOPS NPU, even higher than AMD’s standard 50 TOPS NPU, and part of the collaborative efforts between AMD and HP. It was also stated to be 45% more powerful than Apple’s M4 chips. It’s a bold claim and one without specifics like benchmark tests to back them up, but if it holds remotely true then we could be looking at a truly incredible machine.
Lastly, HP revealed AI software and tools that will be taking advantage of the built-in AI. Zoom, for instance, noticed that 40& of users use fake backgrounds so it improved its image quality while eliminating item disappearance to make for more of a seamless look. Two editing programs, Djay Pro and Blackmagic Design, allow for in-depth depth music and video editing tools that beginners can use. There’s also Loccus.ai, a safety and trust AI tool that uses a Trojan neuro network that combats audio deepfakes, which are some of the most deception and fast-evolving types.
(Image credit: Future)
Accessibility is what AI was made for
The two most promising AI programs I witnessed, were both accessibility-based. Cephable is similar to Google’s own Help Me Hands-Free tool, as it provides face tracking, head tracking, and voice inputs through speech recognition. This allows for alternative navigation methods for those who cannot use a keyboard and trackpad or mouse. Due to this tool running through the AI-powered NPU, it frees up the processor and graphics card for other processes.
OmniBridge is an incredibly useful tool currently in development, targeting an unfortunately niche and underserved market. This program translates sign language into both spoken language and text at the same time, which is made possible through the use of AI analyzing 3D space to interpret gestures in real-time. While the program is still in Alpha, it already knows 5,000 signs, with the ultimate goal of learning 10,000 or full fluency.
It's not hard to see what kind of potential these two programs hold. Not only on their merit but in terms of what other kinds of accessible programs could be created with AI. I would love to find even more of these AI tools developed in the future, it's truly exciting to see what can be done with tech for the greater for once, rather than for solely profit.
(Image credit: Future)
Final thoughts
We’re seeing AI sweeping through the tech industry at an accelerated pace, with every tech giant creating their own AI-based software and hardware to take advantage of the trend. Of course, as with many other recent advancements in technology, it comes at an incredible ecological cost. AI runs out of large data centers needed to power both generative AI and the cloud storage it often relies on, which has been consuming massive amounts of water and other resources.
There’s also the major plagiarism issue, with AI being trained on copyrighted music, papers, written works, research, art and photos, and more. And unfortunately, while laws are being developed to combat this issue, creatives and journalists are still being pushed out of their industries in favor of companies using these AI tools to create content instead.
However, AI has never been all bad, and even now there are ways to apply it to help those in need of these tools the most. From what I saw during HP’s Imagine AI event, the accessibility programs were by far the most creative and helpful of the bunch. They addressed real-world issues and demonstrated why AI is necessary to fix them, using the power of AI to aid underserved markets. Not to mention that this same tech can be run without interfering with other PC processes, meaning you can continue your productivity and creative work or even gaming while having these accessibility tools aid you simultaneously.
If we need to suffer a world that’s oversaturated with AI, with much of it actively harming others and the planet, the only use I want to see is that which genuinely helps others instead. Fewer tools about how to spy on workers using their cell phones in Zoom calls and more tools that translate sign language into multiple spoken languages.
from TechRadar - All the latest technology news https://ift.tt/5fO0Yz4
Full spoilers follow for season 1 ofThe Rings of Power. Potential season 2 spoilers are also discussed.
A new trailer for The Rings of Power season 2 has been forged in the fires of Mount Doom – and, after its worldwide debut at San Diego Comic-Con, it's now available to watch online.
Ahead of the show's second season debuting in late August, the *ahem* powers that be at Prime Video has seen fit to drum up further excitement for The Rings of Power's return with a near-four-minute long teaser that's as action-packed as it is filled with surprises. Believe us, you'll want to watch it immediately – and you can do just that below.
Looks really good, doesn't it? After the somewhat mixed reception for the Amazon TV show's first chapter, it seems season 2 will forge closer ties to what J.R.R. Tolkien wrote about Middle-earth's Second Age than its predecessor. Of course, the odd creative liberty has been taken but, by and large, it looks like some of the biggest moments in The Lord of the Rings' mythology are being brought to life as authentically as possible.
Sure, we'll have to wait for season 2's arrival to see if that's the case, but what does its latest trailer actually tease? Below, I've picked out 14 things you might have missed and/or need explained, and theorized what they could mean for one of the best Prime Video shows' next installment.
1. Sauron returns to Númenor?
Has Sauron returned to the idyllic island of the Edain's descendants?(Image credit: Prime Video)
The very first shot of season 2's new trailer is immediately curious. On the surface, it doesn't appear to be anything of significance – indeed, it simply shows Sauron riding into an undisclosed location while a voiceover from Galadriel plays.
Reading more into it, though, I believe this scene takes place when Sauron voluntarily surrenders to the island of Númenor. His dirty face and somewhat bedraggled clothing suggest he's been in some form of skirmish, while the architecture of the arch he proceeds through is adorned with the Númenórean royal sigil. Indeed, it's identical to the crown worn by Aragorn late in The Return of the King's live-action movie. Oh, and the Númenórean armor, worn by a guard on his left side, is also a dead giveaway.
If I'm correct, I think this scene happens in the season 2 finale, following Sauron's defeat in the War of the Elves and Sauron that, thanks to The Rings of Power season 2's first trailer, we know will be part of this season's plot. So, we might see Sauron – who already visited Númenor in season 1 while masquerading as Halbrand – allow Ar-Phazarôn (more on him later) to take him hostage, thereby enabling Sauron to slowly worm his way into Númenórean high society and set the island nation's downfall in motion. All of that is likely to happen in The Rings of Power's expected third season, but the foundations could certainly be laid in season 2's eighth and final episode – and as part of this particular scene.
2. A helmet-like crown fit for a menacing villain
It looks like Sauron is preparing for a battle - or to rule(Image credit: Prime Video)
The Rings of Power's showrunners have already teased that season 2 will be "all about the villains", so it stands to reason that they feature prominently throughout its latest teaser. That includes clearer looks at secondary foes including fallen elf/orc army commander Adar, as well as the aforementioned Sauron.
We seem to get a glimpse at a prototype of Sauron's iconic armor's headpiece at the 0:10 mark. An orc is carrying what appears to be a crown, but it's shaped similarly to the multi-pronged headgear he wears in the Second Age's final battle – The Last Alliance of Elves and Men, as seen in the prelude to The Fellowship of the Ring film. Is this crown a progenitor, no doubt made by orc smiths, to said helmet? It certainly seems that way to me.
3. Attack of the Barrow-wights
Trailer #2 gives us a much clearer look at one of the show's new spooky creatures(Image credit: Prime Video)
We were given the briefest glimpse of these spooky creatures in season 2's first trailer, but this new reel gives us an even better look at the Barrow-wights who Galadriel, Elrond, and their elven scouting party will duel with at some point.
Judging by their horrifying ghostly demeanor, I definitely wouldn't want to encounter one of these forest-dwelling terrors on a night-time walk. Let's hope these Eriador-based spirits of darkness don't pose too much of a threat to our heroes.
4. Rings, rings, and more rings
We'll see all of the other Rings of Power crafted in season 2(Image credit: Prime Video)
Considering Amazon's Lord of the Rings TV show is called The Rings of Power, you would expect the titular, magic-infused jewelry to show up more often as the series progresses. Thankfully, that'll be the case. Speaking exclusively to TechRadar in October 2022, some of the series' cast teased the creation of the One Ring and its siblings in its sophomore outing.
Based on what we see and hear about them in season 2's official trailer, that's nailed on to happen. We get closer looks at the newly designed elven rings – Narya, Nenya, and Vilya – as well as another good view of King Durin III's blue-gemmed ring. There's also a glimpse of a new Ring of Power that Celebrimbor holds up towards the trailer's end. It seems he's in an underground area, too, which suggests he might be offering it to one of Durin III's fellow dwarf lords.
As for who'll make them, Eregion's elven smiths are seen preparing to craft the rings under the guidance of Annatar, aka Sauron's elven alias. Later, the show's primary villain tells Celebrimbor to "give me the nine", suggesting that the nine Rings of Power he'll eventually 'gift' to the Kings of Men are also going to be created. Evidently, the rings will be just as central to season 2's plot as Sauron is.
5. The Easterlings arrive...
This are either proto-Easterlings or a different group of Rhûn-based warriors(Image credit: Prime Video)
Fans were left disappointed when they learned that the Easterlings, a race of men who hail from the Kingdom of Rhûn, were absent from the Siege of Gondor in the live-action film adaptation of The Return of the King. Indeed, the group was cut from the movie late into its post-production phase in a bid to shave down its theatrical runtime.
Pleasingly, it seems a similar fate won't befall them in The Rings of Power season 2. Between the 1:30 and 1:36 marks, we see a bunch of proto-Easterlings – maybe even each clan's Lord? – featured. One is seen riding their steed amid a sunset backdrop, another is seen spying on Nori and The Stranger (who have traveled to Rhûn to learn more about the latter's past) through a telescope, and the final shot shows five individuals watching someone off-camera.
It's unclear how big of a role they'll play in season 2. But, if this is a tribe (or tribes) of Easterlings, their appearance could be very significant – especially as one of their leaders, Khamul, is one of the nine Kings of Men who receives a ring from Sauron in the source material.
6. ...and the Mystics return
We thought we'd seen the last of the Mystics(Image credit: Prime Video)
Remember when we thought the Mystics all perished in their showdown with The Stranger in The Rings of Power season 1 finale? Yeah, it turns out we were wrong. At the 1:42 mark, we see a group of five new Mystics casting a moth-based spell for reasons unknown. No matter what they're up to, it can't be good news.
7. The Stranger speaks!
Someone's been practicing their A, B, Cs!(Image credit: Prime Video)
It seemed inevitable that The Stranger, who's confirmed as a member of the Istari, aka one of the five wizards sent to help Middle-earth's races defeat Sauron, would continue to be largely mute in season 2. That's based on the footage we'd seen prior to the latest teaser's release.
As it happens, though, we'll finally hear him speak in the show's sophomore season. Indeed, he does as much at the 1:45 mark. The Stranger – who I expect to be revealed as Gandalf at some point – asks Tom Bombadil (more on him in a moment): "Is it my task to stop the fire? Is it my task to face Sauron!?"
8. Welcome to The Rings of Power, 'Old Tom' Bombadil
But I digress. We get a first look at 'Old Tom' in action in this trailer when the enigmatic, jovial, magical individual turns to face The Stranger after the latter asks those aforementioned questions in point seven. He's spotted later in the trailer, too, asking his new friend if he'll abandon Middle-earth's races "to their doom". Expect him to mentor The Stranger throughout this season.
9. Peering into the palantÃr
Well, this isn't good news...(Image credit: Prime Video)
Ar-Pharazôn's rise up the Númenórean ranks looks like it'll continue apace in season 2, judging by the foreboding shot we see of him interacting with one of Middle-earth's eight PalantÃri – indestructible seeing stones that allow for wide-ranging communication across Middle-earth. How and why he's using it – I suspect it's to further his cause to usurp the Númenórean throne – will become clearer when the show returns.
10. Durin's Bane makes its presence felt
You! Shall Not! Pass!... oh, hang on(Image credit: Prime Video)
Ever since The Rings of Power season 1 episode 7 aired, we've known that the Balrog known as Durin's Bane would eventually rouse from its slumber to wage war on the dwarven kingdom of Khazad-dûm. Thanks to co-showrunner Patrick McKay, we've also known that Durin's Bane would next appear in season 2 – McKay telling TechRadar: "Season 2, sir, stay tuned" when asked about when it'll reappear after the show's seventh episode.
Well, McKay wasn't leading me on. Not only did I predict Durin's Bane was responsible for destroying one of the subterranean kingdom's bridges in season 2's first trailer, but it's also set to play a prominent role in the series' next chapter. That's based on the snippets of footage in season 2's newest teaser, which show it escaping its tomb and making a beeline for Khazad-dûm.
11. Water way to go, Miriel
Just taking a quick dip, are you?(Image credit: Prime Video)
Remember that unidentified woman we saw in a body of water, which housed a similarly unknown aquatic beast, in season 2's first trailer? It turns out it was MÃriel, Númenor's Queen Regent. Confirmation of that comes at the 2:15 mark, with the blind monarch entering said lake/sea for reasons I'm not sure of. Oh, and the creature she encounters in both trailers is a Sea Worm. So there, both identity mysteries from the first teaser solved.
12. Enter the Ents
What an Ent-rance!(Image credit: Prime Video)
Arguably the trailer's biggest surprise is saved until the 2:40 mark. A voiceover from someone saying "forgiveness takes an age" begins playing a few seconds earlier, only for the teaser to reveal what appears to be a female Ent – you know, those ancient, slow-moving, sentient tree-like beings – speaking.
We haven't seen an Ent in a live-action Lord of the Rings project since The Return of the King's extended cut, so I couldn't help but smile upon this shock reveal. We get another glimpse at the Ent at the 2:53 mark, which seems to take a swing at a character who could be either Arondir or Isildur. It's hard to make out which individual it is but, given this duo will spend much of season 2 traversing Middle-earth together, it stands to reason that they'll both encounter it.
13. I spy Shelob
"Incy wincy spider..."(Image credit: Prime Video)
Speaking of Isildur, we don't get many glimpses at the Númenórean warrior in season 2's latest trailer. But, when we do get a good look at him towards its climax, we see him fighting a spiderling in a dark cave.
Now, you might be thinking this is just an ordinary arachnid, but I can reveal the eight-legged freak we see is actually Shelob. That's right, the giant spider that Frodo and Sam encounter on their long trek Mordor in The Return of the King will appear in The Rings of Power season 2 – an Amazon press blast accompanying its newest trailer confirming the monstrous creepy-crawly's cameo.
14. They have a Hill-troll
Keep trollin', trollin', trollin'...(Image credit: Prime Video)
There's one final monstrous reveal before the trailer ends, too. At the 3:14 mark, we catch a fleeting glimpse of a new troll, which as part of Sauron and Adar's massive orc army will play a huge role in the Siege of Eregion – likely to be season 2's tentpole battle.
For those who missed it at the time: this creature was first revealed by Empire magazine on July 1, with the entertainment publication confirming it'll be named Damrod. Oh, and it'll be a Hill-troll that's referred to as the "eater of dragon bones". A colossal being not to be trifled with, then!
The new SearchGPT feature introduced by OpenAI has stumbled in its rollout, as first discovered by The Atlantic. SearchGPT is a new tool designed to combine OpenAI's AI models with real-time web data for faster, more accurate answers. It's not widely available but represents OpenAI's vision of how AI and search will augment each other in the future.
In a prerecorded video showcasing SearchGPT, the tool provided incorrect information about the dates of the Appalachian Summer Festival in Boone, North Carolina. The July 29 to August 16 dates SearchGPT reported are far from the June 29 to July 27 dates the event actually occurs. As reported by The Atlantic and further confirmed, the dates of the festival in SearchGPT's response are when the box office is closed. You can watch the full demo here.
SearchGPT Still Looking
AI hallucinations and errors are perennial complaints, a universal issue pretty much every AI user has encountered. In that sense, the mistake is not a huge deal. However, the result here somewhat undermines OpenAI's pitch for SearchGPT. With reliability and transparency at the heart of SearchGPT, according to OpenAI, it stands out when a hallucination like this still happens.
This isn't an issue that is unique to OpenAI, of course. You may recall the embarrassing errors made when Google debuted its AI assistant Bard (now Gemini). In that case, a live demo claimed the James Webb Space Telescope took the first pictures of a planet outside the solar system, even though they were actually taken by the European Very Large Telescope. It was so bad that some people attributed a subsequent $100 billion dip in Google stock price to the gaffe. Arguably, having the error in a video that OpenAI could have redone or edited differently at any time is worse than a real-time misstep.
Regardless of the mistake, OpenAI will likely not slow down in developing AI search. The demand for accurate, speedy answers to questions about things happening right now has already propelled AI integration into existing search engines, including Google and Microsoft Bing. It's also the impetus behind search-centered generative AI chatbots like You.com and Perplexity.
But, if OpenAI can release its own option, especially tied to ChatGPT, it will, at the very least, score a robust slice of the search AI market. That's predicated on getting people to trust SearchGPT to provide accurate answers, of course. You could ask SearchGPT to give the odds of that happening, but you might want to follow up on your own to confirm what it says.
Orico, in partnership with Western Digital, has launched a Kickstarter campaign for the Data Matrix Ultra, a new storage solution designed for creative professionals. This system looks well suited to handling high-resolution 4K and 8K video and supports intensive post-production workflows.
The Data Matrix Ultra is designed to evoke the aesthetics of Apple products. It combines fast M.2 SSDs (with speeds up to 3000MB/s) with high-capacity WD Red HDDs, and includes CFexpress Type-B slots capable of speeds up to 1500MB/s, which is fast enough to edit 8K video.
This storage system comes in a variety of capacities ranging from 5TB to 112TB, with options for future expansion and upgrades to meet growing project demands. It allows for the daisy-chaining of up to five devices, for a potential maximum of 560TB of storage. Data Matrix Ultra supports a variety of RAID configurations - RAID 0, 1, 5, 10, and JBOD - providing data security through redundancy and backup options.
Already funded
The Data Matrix Ultra also focuses on minimizing operational noise with features such as anti-vibration mounts and magnetic levitation fans. Orico says this will help with a quieter workspace, crucial for creative professionals working in sound-sensitive environments.
For me, the Data Matrix Ultra brings to mind the OWC ThunderBay Flex 8 storage solution, one of the fastest and most flexible storage options available which also has an Apple inspired design (it looks not unlike the large Mac Pro). OWC’s solution offers 18TB to 172TB of storage and is priced at $2,479.00. Backing prices for the Data Matrix Ultra start at $699 for a 1 bay 5TB model, a 36% discount off the $1,099 MSRP, rising up to $4,999 for the 112TB 5 bay model, marking a 62% reduction from the eye watering $12,999 MSRP.
Backing crowdfunded products obviously comes with risks. Orico notes “the high price of the Data Matrix mini could limit the number of potential backers, especially in a competitive market where consumers might choose more affordable alternatives”. That said, the Data Matrix Ultra has already found an audience, amassing $46,885 of its $5,000 goal (via 28 backers), with over two weeks remaining in the campaign.
Cybersecurity researchers from ESET have warned of a recently-discovered vulnerability in the Android version of the popular instant messaging application Telegram.
The vulnerability allowed threat actors to deploy malware on the vulnerable devices, and apparently - it was being actively exploited for weeks.
A threat actor called Ancryno took to a Russian-speaking underground forum in early June 2024, to sell a zero-day exploit for Telegram versions 10.14.4 and older. This drew the attention of ESET’s experts, and when a proof-of-concept (PoC) was published, they picked up the malicious payload, analyzed it, and confirmed that it works.
Fake prompts
The vulnerability allowed threat actors to create malicious .APK files (Android installation packages) which, to the recipient, look like a video message. Since Telegram automatically downloads all multimedia, all the victim needs to do is open up the chat window to receive the payload.
Users who disabled the automatic download of multimedia files need to tap on the received message once to trigger the download.
This leaves the problem of actually running the file, since the APK still needs to be installed. The hackers partially solved it by displaying a fake prompt that the video needs to be played in an external player. Accepting this prompt triggers another one which says that Telegram is barred from installing APK files. If the victim ignores all of these red flags, they will end up with the installed malware.
Further analyzing the threat actor’s infrastructure, ESET found two malicious payloads hosted online, one that pretends to be Avast Antivirus, and a fake “premium mod” for xHamster (a website with adult content).
The researchers reported their findings to Telegram’s developers, which came back with a patch on July 11. In its writeup, BleepingComputer points that the flaw was running wild for at least five weeks, giving crooks plenty of time to target Telegram users.
The earliest patched version is v10.14.5. Telegram’s desktop app was never vulnerable.
Cybersecurity researchers from Trend Micro recently found a Linux variant of the dreaded Play ransomware strain targeting VMWare ESXi environments.
In a technical breakdown, Trend Micro’s Threat Hunting team said this was the first time Play was seen targeting ESXi environments, and it could be that the criminals are broadening their attacks across the Linux platform, giving them an expanded victim pool and more successful ransom negotiations.
Play was first spotted more than two years ago, and since then it became popular for its double-extortion tactics, evasion techniques, custom-built tools, and a “substantial impact” on companies in Latin America, the researchers explained.
Prolific Puma and Revolver Rabbit
Businesses usually use VMWare’s ESXi instances for virtual machines, where they host critical applications, data, and integrated backup solutions. By targeting these endpoints, Play’s operators could reduce the chances of the victim recovering any encrypted data. Therefore, their negotiation position becomes that much better. Besides going after Linux endpoints, the new variant was also able to successfully evade security detections, Trend Micro added.
Analyzing the infrastructure used for these campaigns, the researchers found a peculiarity - the URL used to host the encryptor is related to a threat actor known as Prolific Puma. This group is known for offering URL-shortening services to criminals, making phishing attacks more convincing, and thus, more disruptive.
In late 2023, researchers Infoblox discovered a major link-shortening operation in which the criminals would use a registered domain generation algorithm (RDGA) to create domain names in bulk. Then, they would use those domains to provide a link-shortening service to other malicious actors.
Earlier this month, the same company found a threat actor called Revolver Rabbit using RDGAs to register more than 500,000 domains, an effort on which they spent more than a million dollars. The hacker used the RDGA to create command and control (C2) and decoy domains for the XLoader infostealing malware.
Sony has long been a key player in digital storage technologies. The company introduced the CD in 1982, which it developed in collaboration with Philips, revolutionizing how data and music were stored and accessed. In 1995, Sony launched the Digital Versatile Disc (DVD), a joint effort with Philips, Toshiba, and Panasonic. This was followed by the Blu-ray Disc in 2006, which offered even greater storage capacity suitable for high-definition video and data.
In 2014, the company, in partnership with Panasonic, came up with a new optical storage format for professionals, the Archival Disc, which could store a whopping 300GB of data.
The writing has been on the wall for Sony’s older optical solutions for a while however, and the company has said it plans to phase out its recordable disc production, which includes Blu-ays, having previously announced plans to cut a significant number of jobs at its optical media plant in northeast Japan.
Alternatives step up
"The growth of the cold storage market has not reached our expectations, and the performance of the storage media business as a whole continues to be in the red," a Sony Group spokesperson said. "We have determined that it is necessary to review the business structure to improve profitability."
Another casualty of this move will be Sony’s Optical Disc Archive Cartridges. Announced in 2020, Generation 3 of the product has a 5.5TB capacity courtesy of a double-sided disc with triple layers per side for write-once recording. This latest incarnation (and last, as it happens) offers a much greater capacity than Generation 2, released in 2016 (3.3TB), and the initial release, introduced in 2013, which came in various capacities up to 1.5TB.
Sony’s move will mostly impact consumers as the company says it will continue to sell B2B products from existing supplies. It has produced enough to meet demand for quite some time, but if you rely on optical storage you might want to stock up now.
A number of tech companies are working on alternative forms of storage that could fill the void that optical technology’s decline will leave. These include ceramic, Silica (quartz glass), and DNA-based storage.
OpenAI has added more business-focused controls to its Enterprise subscription for ChatGPT in a bid to help tackle companies’ biggest concerns.
The updates focus on compliance, data security and scaling access to generative AI while also tackling the broader issue of meeting various stringent requirements across regulated industries globally.
Key to the update is the Enterprise Compliance API, which helps ensure alignment with standards like GDPR, HIPAA and FINRA.
ChatGPT just got even more Enterprise controls
The Enterprise Compliance API provides users with a detailed and time-stamped record of all interactions, including conversations, uploaded files and metadata, which can be used for auditing.
OpenAI also introduced third-party integrations with Forcepoint, Global Relay, Microsoft Purview, Netskope, Palo Alto Networks, Relativity, Smarsh and Zscaler, which can be used for things like data archiving, audit trails, redaction, retention and policy enforcement. The San Francisco-based organization committed to working with more partners in the future, hinting at the expansion of its support for third parties.
As well as compliance features, OpenAI will also roll out an identity management system to make managing users easier for IT admins. The System for Cross-domain Identity Management, currently in beta, will work with “most company directories,” including Okta Workforce, Microsoft Entra ID, Google Workspace, and Ping.
The final control upgrade for ChatGPT Enterprise customers is more granular control over custom GPTs. Admins will now be able to create lists of approved domains, rather than completely allow or block the use of GPTs.
In an expression of commitment to privacy and security, the company affirmed that no customer data or metadata is used for training models, and that data is encrypted both at rest and in transit.
The business-focused subscription is now believed to boast more than 600,000 users, with cheaper subscriptions for SMBs and consumers also available for monthly fees.
Choosing the best photos for your dating profile can be tough, so Tinder created a virtual curator to help you out. The artificial intelligence-powered Photo Selector will analyze your shortlist of potential photos directly from your phone and suggest those its model predicts would present you at your best.
To use the feature, a Tinder user takes a selfie for the AI to know what they look like, then permits the app to look at photos on their phone. The AI model picks out images for the user to review and decide whether to add them to their profile. Photo Selector is coming first to Tinder users in the U.S. this month, with international rollouts later this summer.
Tinder hopes the AI tool will smooth the path for setting up a dating profile. According to its own Online Opinion survey of young single people, 52% struggle to pick a profile image for dating apps, and single people under 25 spend 33 minutes on average picking a photo for their dating app profile. Perhaps it’s unsurprising then that 68% of them welcome the idea of AI assistance in picking their photos.
Though Tinder doesn’t outright say so, the study did suggest straight men in particular need the help. Straight single women find profiles with at least four images highlighting a man’s personality more attractive, and more than one face photo ups a guy’s chance of matching by 71%.
“We’re proud to be the first dating app to roll out an AI tool that can make the profile-building experience significantly easier -- an area we know is one of the hardest parts of dating,” Tinder CEO Faye Iosotaluno said in a statement. “As the category leader, we’re pushing ourselves to define the industry’s best use cases for meaningful consumer AI integrations,”
Discreet AI
Tinder didn’t make much of any privacy issues with regard to the new feature, even though access and use of personal photos with AI models may make some nervous. Users might inadvertently expose sensitive or private images by granting the app access to their camera rolls. The company has data protection policies and security measures in place, but when it comes to something as intimate as photos for dating apps, it’s easy to imagine some hesitating without enough transparency and trust regarding how the images are used, stored, and protected.
That’s on top of the facial recognition element. While it is necessary to curate the photos successfully, the biometric data involved is arguably even more sensitive. Tinder may have to make an extra effort to assure users their data is anonymized when the AI processes it and that it’s not shared with third parties. Still, as AI assistants and related tools become more ubiquitous, ones that help people’s online profiles pop, whether on dating apps or anywhere else, will likely become very common.
“As demonstrated by our Photo Selector feature, we’re developing AI tech to assist you in making decisions, not to make them for you,” Iosotaluno said.
Back in 2021, Samsung announced the industry’s first High-K Metal Gate (HKMG) based DDR5 memory, the use of which doubled the speed of DDR4 while reducing energy use by 13%, with Samsung’s DDR5 stacking eight layers of 16Gb DRAM chips to offer a whopping 512GB capacity.
Fast forward to 2024, and at Computex, ServeTheHome came across the Gigabyte R283-ZK0 motherboard, shown above. Integrating the maximum number of DIMM slots into servers, specifically those using modern configurations with 12 DDR5 channels per CPU socket and two DIMMs per channel (totaling 24 DIMMs per socket), presents significant challenges in terms of both ensuring the DIMMs function correctly together and managing the physical space required to fit all these components inside the server chassis. This is a problem the Gigabyte R283-ZK0 motherboard skillfully addresses.
This powerhouse is not just another server board, it packs in 48 artfully arranged DDR5 memory slots, allowing for up to 24TB of RAM when equipped with Samsung's 512GB DDR5 modules.
Twisted slots
From the outside, the R283-ZK0 resembles a typical 2U server with its 12x 2.5-inch hot-swap NVMe/SATA/SAS drive bays and a large fan wall, but the magic is revealed once you remove the lid. Inside, the server reveals a CPU slot positioned at the back, just in front of some PCIe risers. But the real marvel is above the system where the motherboard stretches out with an impressive array of 24 DDR5 RDIMM slots per CPU - totaling 48 slots in a dual second-gen AMD EPYC 9004 series setup.
As ServeTheHome’s Cliff Robinson says, "To fit this many DIMMs, the CPUs cannot be aligned next to one another. Instead, there are staggered sets of 6-2-4-4-2-2-4 DIMM slots. The slots are twisted on the motherboard at an angle to allow for some overlap of the slots and sockets." It's an impressive piece of engineering for sure.
Other features of the Gigabyte R283-ZK0 include dual 10Gb/s LAN ports (Intel X710-AT2), one PCIe Gen5 x16 FHFL (full-height, full-length) slot, and one OCP 3.0 Gen5 x16 LAN mezzanine slot for use with network cards.
The best sports movies aren't just about sports. Take the recent movie of the day Rocky for example: it's about working class pride, about love and about overcoming adversity, so it's more than just a boxing movie. And the same applies to The Long Game, because while it's an uplifting underdog story set around the game of golf it's also about some of the toxic attitudes prevalent in the 1950s.
The Long Game is based on the true story of five Mexican American high school students who were banned from playing golf in the 1950s by an all-white Texan country club. Refusing to take no for an answer, they built their own course, formed their own team, and became state champions.
As Variety put it: "Winner of an audience award at the 2023 SXSW film fest, The Long Game is an utterly predictable yet thoroughly engaging period drama." It's "appealingly told with sincerity and taste," says RogerEbert.com, while the Hollywood Outsider says that "The Long Game is an emotional crowd-pleaser".
While few critics can fault the heart of the film, many felt that it could have been more focused: Collider said that "while the book this film is based on, Mustang Miracle by Humberto G. Garcia, certainly documents an impressive story of an underdog overcoming odds, the film doesn't quite reach the heights it's trying to... It juggles too many things and ends up dropping the ball." While the movie does address racism head-on, "it struggles to find balance and nuance along the way".
The result is "a gently rousing family-friendly drama," The Hollywood Reporter says. It "could have used more topspin, but [it's] a winning round nonetheless", which is why it makes the cut for our best Netflix movies list.
Vaio has announced the Vision+ 14 portable monitor as part of its 10th anniversary celebrations.
The Vision+ 14 weighs just 325g, equivalent to 0.72 pounds, and features a 14-inch, 1920x1200, 16:10 aspect ratio display. The screen boasts properties such as 400 nits of brightness, wide viewing angles, anti-glare technology, and a wide color gamut.
The monitor isn’t just light; it’s also incredibly slight - it measures just 3.9mm at its thinnest part. The back panel of the main unit is made of 3D molded carbon, making it both lightweight and strong. It comes with a solid metal kickstand for easy use alongside PCs or laptops, particularly in compact workspaces, and comes with a cover stand.
Extra screen for laptops
The monitor, which Vaio is proud to stress is made entirely in Japan, comes with dual USB Type-C ports that support PD pass-through and DP-Alt. It is priced at 54,800 yen, approximately $340 including sales tax. There are two versions - one for consumers and another for businesses. There doesn’t seem to be any difference in the models however, just the option for firms to bulk buy at reduced rates.
We’ve reported on a couple of laptops with dual screens in recent months, including the GPD Duo and the Acemagic X1, suggesting it could be a trend for 2024. If you like the idea of a laptop with twin screens stacked above each other, but don’t want to spend money on a new PC, the Vision+ 14 could be the percent solution as the included cover stand will allow it to sit above your existing screen.
Vaio says the Vision+ 14's thin and light design addresses the issue highlighted in a company survey that found 50% of people said a sub-display is a necessary device for hybrid work, and of those, over 40% said they had issues with portability.
Once a division of Sony, Vaio was sold off in 2014 when the PlayStation maker decided to exit the PC market. Since then, Vaio has continued to make products in Japan and has recently begun exporting its line of computers to markets like the US and Brazil.
Many mini PCs are becoming increasingly popular because they offer the power of a regular computer in a much smaller package, making them great for people who don't have a lot of space or who need a portable solution that they can easily take with them on the go.
One of the more unusual mini PCs we've come across lately is the Meenhong JX5, which stands out because it looks like a smartphone and runs Windows 11, bringing to mind Microsoft’s failed Windows Phone experiment.
The JX5 has a sturdy aluminum body and a 5.5-inch touchscreen with FullHD (1920x1080) resolution. The screen is big enough to be usable, but you may struggle to read smaller text or interact with some on-screen elements. At 161mm tall and 92mm wide, it’s a little larger than an iPhone 15 Pro Max, but considerably thicker. A chunk of the 23mm depth is down to the inclusion of a cooling fan.
Limited battery life
It’s another mini PC powered by an Intel N100 processor with Intel UHD Graphics. It comes with up to 12GB LPDDR5 4800MHz soldered-in RAM and you can buy it on AliExpress where it goes under the SZBOX brand. The barebone model with no SSD and no OS sells for $220. If you opt for the top version, with a 2TB SSD, the price rises to $356.32.
The Meenhong JX5 has WiFi 6 and Bluetooth 5.2, a full USB Type-C port, a USB-A 3.0 port, two USB-A 2.0 ports, HDMI 2.0, Gigabit Ethernet, and a 3.5 mm audio jack. There’s no SD card slot, though. It’s able to power two additional screens at 4K@60fps, so you can use it for a three-screen setup.
We’ve covered its predecessor, the Meenhong JX2, before, and while that device was intriguing, the lack of a battery massively limited its portability. The JX5 does have a battery, but it’s a 1500 mAh model, so you’re going to be looking at about an hour’s use between charges. Like the JX2, it doesn’t have a webcam either, which seems like a strange omission. Although the JX5 runs Windows 11, you can opt for a Linux OS if that better suits your needs.
This mini PC brings to mind similar products like HiGOLE's tablets, the GOLE1 and GOLE1 Plus, both of which we've reviewed here before.
