According to Microsoft, this exact issue is affecting certain installs of Outlook, Word and OneNote, and its Outlook and Office teams are currently investigating (via Neowin). In a support post, MS defines the problem as:

"After updating to [Office] Version 2407 Build 17830.20138 or higher you find that Outlook, Word, or OneNote may unexpectedly close when typing or doing other authoring tasks such as spell check."

Yep, the simple act of typing or performing a spell check can cause the applications to close, undoubtedly inspiring a choice selection of curse words from affected users. Microsoft says that you can confirm the issue by looking for Event 1000 or Event 1001 in the Windows Event Viewer Application Log and that the issue may be caused by older language packs:

"The faulting module name will vary depending on what language packs you have installed. For example, mscss7it.dll for Italian, or mscss7ge.dll, for German, and others could include: EN, ES, FR, GE, IT, NP, PB."

Currently, MS advises a workaround involving an online repair of the affected Office application, searching for and uninstalling old Office language installations, and reinstalling the Language Accessory Pack for Microsoft 365. 

Affected users are also pointed towards a thread on the Microsoft forums in which multiple users report their Outlook installations crashing from simply typing, saving drafts, and typing in different languages—both in the Windows version and the Android application. 

Many of the users in the thread report that they were typing in (or spellchecking) German when the problem occurred, although by the look of the support post MS seems to think multiple languages could cause the affected apps to close.

Chalk one up for Notepad, I guess, although it's not like you can send an email with the app I internally refer to as "Ol' Reliable". Still, it does have a surprisingly good spellcheck and multiple useful features these days, and I've yet to have it crash on me while I'm taking notes. Sometimes the classics are the best, ey?

There might be some PTSD involved here. Once upon a time I owned a MacBook Air, and I loved that laptop. Brilliant piece of hardware: no Windows laptop has ever had a trackpad that good. It weighed nothing. The battery life was killer in its day. But man there were some things about Mac OS that I couldn't stand, including particularly Apple's preference for hiding away all my files, as if catching sight of ".jpg" would immediately stricken me with the bubonic plague. If I wanted to look at the pictures from my phone in the Photos app, easy peasy, but finding the actual files involved spelunking through some horrific sequence of folders.

One of my strongest opinions about computers is that I should not have to dig through, like, "User > Wes > .temp > astuvkcaqcf > 4748949585" just to find some damn files. 

Maybe Apple's gotten better about that in the years since—I have no idea. But to see Valve go down the same road now makes me want to pick up my PC and walk directly into the sea with it. I live about two miles away from the ocean which is a long time to carry a 30-ish pound desktop, but on the bright side, holding onto it will ensure I sink to the bottom of the sea where Steam's screenshot interface can no longer hurt me.

For many years the Screenshot window has included an all-powerful button, a simple vehicle for raw, efficient digital scrapbooking. You clicked it, and it opened up a Windows Explorer window straight to all the screenshots you'd ever taken for that individual game using Steam. Easy! Its other features have always been far less useful to me: I rarely want to upload a screenshot to the Steam Cloud because I do not post on the Steam forums. I do not need to use Steam's screenshot management window because if I'm taking a screenshot, I just want the dang .jpg, or a whole handful of them, to upload onto this website. 

Now in Steam's latest beta, clearly pulling in a simplified interface that prioritizes the Steam Deck, the ability to browse to one-click jump to that folder is gone, replaced by a Share button. You also used to be able to right-click an individual image and jump to the file location that way. That's gone, too. Valve wants you to stay inside Steam for all this stuff now, furthering a trend where all computing happens in an app and barely acknowledges the computer it's installed on.

Has anything been more of a scourge on computer interfaces this last decade than the Share button? It's the ultimate "clicking this doesn't actually do shit, but it does open up a list of other things you can do that used to have their own buttons" bit of obfuscating UI design. Via the Share button, Steam now lets me, one at a time, click "Save image" and choose where to save a screenshot on my computer.

But it's already on my computer, Valve. Where is it? Where did you put it? WHERE ARE MY FILES!!

The old interface remains in the non-beta branch of Steam for now, but I must implore Valve: bring back your second-best button. Don't make Steam another casualty of the obfuscation era of interface design. Sometimes The Old Ways are best. I would even accept a return to skeuomorphism if that was the only way to get back the simple click action I crave. Animate your screenshot window rolling open like a file cabinet full of manila folders if that's what it takes. Just give me back my button!

Footage of the robot in action was posted on X by Google DeepMind and as someone who has some experience of programming robot arms, albeit in a limited manufacturing engineering capacity, I was really impressed by what I saw, and the study's research paper is well worth reading.

It's more than just a camera keeping track of the ball and then motors wildly swinging a paddle about. To start things off, Google DeepMind built a database of all the initial states a table tennis ball could have, such as position, speed and spin. From here, the robot arm just practised various movements, getting used to switching between fore- and backhand grips, applying topspin, and so on.

Then it was pitched against real players, with the AI system designed so that would monitor how different people would behave and play, and use that information to refine the overall algorithm. Its success rate was tracked and the selected strategy self-adjusting in real-time accordingly.

Google DeepMind says that the robot played against 29 opponents, ranked into four different skill levels, and after completing all the matches, it came roughly in the middle of them—essentially the same as an "intermediate amateur."

Of course, it got its metal butt well and truly handed to itself by the better players and one clip shows a person starting slowly but then trouncing the AI with one quick shot. Google DeepMind says that this is to be expected, as aspects such as how paddle rubber affects spin are difficult to model properly in simulations.

But even so, it's still an impressive achievement, and I can think of multiple applications that future versions of the technology could be used for. Mass-manufacturing production lines that use robots for things like painting or welding are brilliant but often struggle to cope with minor mispositions or changes in lighting. An AI system trained to react properly to all of these will help prevent such issues from stalling the production line.

And I can foresee a time when seriously injured people could be routinely relying on AI-trained robot arms to handle things for them, while they recover and return to full strength and mobility.

For now, though, getting better at table tennis is where it's at—and I'm 100% on board with that goal. After all, it's infinitely better than getting better at robotic nuclear armageddon.

It might seem a little hard to believe but Microsoft's Paint app has been part and parcel of every version of Windows, making it almost four decades old. However, in 2017 with the Windows 10 Creators Update, Microsoft thought we could all do with a better version and gave us Paint 3D, with the main feature set being its support for 3D models and rendering.

But as Windows Central reports, despite having some genuinely good features, time is being called on the seven-year-old app and after November 4 2024, you won't be able to download it from Microsoft Store, nor will it ever receive any future updates.

To be honest, the news shouldn't come as any surprise, especially given that Microsoft itself has been rather keen on Paint of late, updating it with a background removal tool and a generative AI system last year, to name just a few of the recent changes.

Like many of my colleagues in the PC Gamer hardware team, I have a genuine fondness for Paint and routinely use it to do basic annotations, image format changes, resizing, and so on. For more complex work, I typically use GIMP (occasionally Photoshop) and when I worked in the field of engineering, anything 3D-related would involve firing up AutoCAD, Inventor, or Fusion 360.

What I've never needed was something that was only a piecemeal version of all of these tools and given Paint 3D's swift consignment to software history, I strongly suspect I'm not the only one.

Windows 10 might still be the world's most popular Microsoft operating system but that hasn't saved the likes of Cortana and the Tips tool being quietly booted into netherspace. At least Paint 3D can join them with a sense of being quite useful and I wouldn't be surprised if there are some folks out there who will be sad to see it go.

On an average day about a dozen new games are released on Steam. And while we think that's a good thing, it can be understandably hard to keep up with. Potentially exciting gems are sure to be lost in the deluge of new things to play unless you sort through every single game that is released on Steam. So that’s exactly what we’ve done. If nothing catches your fancy this week, we've gathered the best PC games you can play right now and a running list of the 2024 games that are launching this year. 

Burn

Steam‌ ‌page‌ ‌
Release:‌ August 9
Developer:‌ Convict Games

This visual novel traces the adventures and misadventures of Nina Burn, a globally popular Finnish pop star with three albums under her belt. Where next for Burn? Will she fade into irrelevance, reassert her dominance, or "become a monster"? There are sixteen endings here, and so plenty of answers to that question. Based on her logo, maybe there's a pop / black metal hybrid in Burn's waters? This game is made by the team responsible for Stone, which is an "Aussie stoner noir story" game featuring a koala. There's a demo for Burn, in case you're not 100% sure you want to deliberate over the career moves of a mid-career pop star (which seems weird to me). 

Vladik Brutal

Steam‌ ‌page‌ ‌
Release:‌ August 10
Developer:‌ Narko Games

This solo developed first-person shooter stands out because it seems heavily inspired by Half-Life rather than the usual close-to-zero narrative boomer shooters. Set in an Eastern European dictatorship, its art style is definitely reminiscent of Valve's classic, though Vladik Brutal leans in more on Soviet-era architecture and sooty urban environments. Most importantly, the movement and combat looks impeccably smooth here, with production quality that could rival bigger indie studios. It's definitely worth a look if you're one of those people decrying the death of cinematic singleplayer FPS games.

Cat Quest III

Steam‌ ‌page‌ ‌
Release:‌ August 8
Developers:‌ The Gentlebros

I played Cat Quest 2 with my seven-year-old son and it went down a treat. These games aren't super complex takes on the action RPG, so they're great to play with people inexperienced or otherwise indifferent to the genre. As far as I can tell, Cat Quest 3 is more Cat Quest, except now there are boats, pirates, and some deviations into 2D sidelong navigation. If you crank the difficulty up you're in for a challenging time, of course, but Cat Quest is basically a cosy game with, uh, lots and lots of blood-free killing. 

Volgarr the Viking II

Steam‌ ‌page‌ ‌
Release:‌ August 7
Developer:‌ Crazy Viking Studios, Digital Eclipse

I hated the first Volgarr the Viking, in the sense that I loved it but it kicked my arse so hard I felt much, much older after finishing it. So I'm determined not to touch this sequel to the tough as nails 2D platformer. I need to be careful of my heart rate. It looks to be a sequel in the most common sense: it's similar to the first game just with more stuff. More power-ups, more magic, more baddies. I hope you have huge reserves of patience. I feel angry just thinking about it.

Five Nights at Freddy's: Into the Pit

Steam‌ ‌page‌ ‌
Release:‌ August 8
Developer:‌ Mega Cat Studios

I doubt many FNAF fanatics "missed" this game, but it's among the few in this universe that I've been tempted to try for myself, so I'm including it for everyone else. It's a 2D pixel art take on the horror series with puzzles and tense pursuits. It's developed by the studio responsible for WrestleQuest, and its pixel art is similarly colourful and chunky.

You could imagine a game dev using a chatbot to help gamers ask questions about everything from how to complete a game to applying the best settings and fixing technical issues. There is, inevitably, a catch, however.

According to Zenity, an AI security specialist, Copilot Studio and the chatbots it creates are a security nightmare (via The Register). Zenity CTO Michael Bargury hosted a recent session at the Black Hat security conference, digging into the horrors that unfold if you allow Copilot access to data to create a chatbot.

Apparently, it's all down to Copilot Studio's default security settings, which are reportedly inadequate. Put another way, the danger is that you use that super-easy Copilot Studio tool to create a super-useful tool that customers or employees can use to query using natural language, only to find it opens up a great big door to exploits.

Bargury demonstrated how a bad actor can place malicious code in a harmless-looking email, instruct the Copilot bot to "inspect" it, and—presto—malicious code injection achieved. 

Another example involved Copilot feeding users a fake Microsoft login page where the victim's credentials would be harvested, all displayed within the Copilot chatbot itself (via TechTarget).

Moreover, Zenity claims the average large enterprise in the US already has 3,000 such bots up and running. Scarily, it claims 63% of them are are discoverable online. If true, that means your average Fortune 500 outfit has about 2,000 bots ready and willing to spew out critical, confidential corporate information.

"We scanned the internet and found tens of thousands of these bots," Bargury said. He says Copilot Studio's original default settings automatically published bots to the web without any need to authenticate to access them. That's since been fixed after Zenity flagged the problem to Microsoft, but it doesn't help with any bot built before the update.

"There's a fundamental issue here," Bargury says. "When you give AI access to data, that data is now an attack surface for prompt injection." In short, Bargury is says that publicly accessible chatbots are inherently insecure.

Broadly, there are two problems here. On the one hand, the bots need a certain level of autonomy and flexibility to be useful. That's hard to fix. The other issue is what seems to be some fairly obvious oversights by Microsoft.

That latter issue perhaps shouldn't be surprising given the debacle over the Windows Copilot Recall feature, which involved taking constant screenshots of user activity and then storing them with essentially no protection.

As for what Microsoft says about all this, it provided a slightly salty response to the Register.

"We appreciate the work of Michael Bargury in identifying and responsibly reporting these techniques through a coordinated disclosure. We are investigating these reports and are continuously improving our systems to proactively identify and mitigate these types of threats and help keep customers protected.

"Similar to other post-compromise techniques, these methods require prior compromise of a system or social engineering. Microsoft Security provides a robust suite of protection that customers can use to address these risks, and we’re committed to continuing to improve our safety mechanisms as this technology continues to evolve."

Like so many things with AI, it seems security is another area that will be a minefield of unintended consequences and collateral damage. It does rather feel like we're an awfully long way from the prospect of safe, reliable AI that does what we want, and only what we want.

Oligo Security's research team has discovered a “0.0.0.0 Day” vulnerability that affects Google Chrome/Chromium, Mozilla Firefox and Apple Safari browsers, enabling websites to communicate with software running on MacOS and Linux systems (via The Hacker News).

The vulnerability means public websites using .com domains are able to communicate with services running on the local network by using the IP address 0.0.0.0 instead of localhost/127.0.0.1.

The good news, if you're a Windows user at least, is that Microsoft's OS blocks 0.0.0.0 at a system level. Hooray for the sometimes-rarer-than-we'd-like Microsoft security win. The bad news for the rest of you is that this loophole is said to have been exploitable since 2006, which means it has been an active cybersecurity vulnerability for an astonishing 18 years.

It's said that the percentage of websites that communicate using 0.0.0.0 is on the rise. Looking at Chromium counters, Oligo has identified 0.015% of websites that could potentially be malicious. That might not sound like a lot, but according to the team, there are an estimated 200 million active websites as of August 2024. 

That's potentially 100,000 websites communicating over that particular IP address, although how many of them are using that capability for nefarious purposes is currently unknown.

Oligo disclosed its findings to security teams from each of the major browsers affected in April 2024, which the company says was acknowledged by each, and that changes are underway to plug the vulnerability.

However, it's up to browser developers to implement their respective fixes, and those fixes have been rolling out to different browsers at different times.  Chrome is already blocking access to 0.0.0.0—starting with Chromium 128—and Google plans to gradually roll out the change with completion set for Chrome 133.

Apple-based browsers like Safari use Webkit, which has already blocked 0.0.0.0. since the report. As for Mozilla Firefox, there is currently no immediate fix, but Mozilla has changed the Fetch specification to block 0.0.0.0 attempts. According to Oligi, "at an undetermined point in the future, 0.0.0.0 will be blocked by Firefox."

Call me slightly smug, but given some high-profile Windows cybersecurity-related failures of late I'll take any win I can get. If you're a Windows PC user, it's finally time to take a victory lap. This one's not on us, folks, and we can rest easy in our beds tonight.

DoctorRofatnik created Sonic Eclipse Online, a game where Roblox users could race as Sonic and friends, and it proved unsurprisingly popular with the young audience. As is typical of the platform's experiences, users could buy costumes and trinkets with Robux, and DoctorRofatnik claimed to be one of the best-paid creators on the platform. He further claimed to be Jadon Shedletsky, the younger brother of Roblox's longtime creative director John Shedletsky.

Many users were attracted by DoctorRofatnik's story, popularity and position on the platform: even his overtly sexual sense of humor. Others smelled danger, and perhaps the most shocking element of this story is that Roblox users seemed more alert to the danger posed by this individual, and did more to expose his behaviour and even identify him, than Roblox and the conventional authorities.

In late 2020 screenshots were posted to Twitter showing a private chat DoctorRofatnik had with a 12 year-old, in which he wrote: "soon I’ll corrupt you beyond your wildest dreams. Words cannot explain what I want to do with you. You’re the reason why I’m gonna end up behind bars."

One of the figures involved in exposing DoctorRofatnik is Ben Simon, a controversial Roblox YouTuber who goes by the handle Ruben Sim. He was sent the above screenshots of the chat exchange and, as well as posting them to Twitter, made a video exposing the behaviour that he also sent to Roblox's developer relations team. Other users reported the account, as did the mother of the girl DoctorRofatnik had messaged. 

Roblox banned the DoctorRofatnik account, and reported it to the National Center for Missing & Exploited Children. He nevertheless remained active on other accounts, and boasted the controversy had boosted sales in Sonic Eclipse (which remained available). Simon thought the Roblox response was weaksauce: "Roblox spends so much time, effort and money convincing parents that their platform is safer than it actually is."

Further tips would lead Simon to DoctorRofatnik's real identity: Arnold Castillo of Paterson, New Jersey. He called the Tucson police, who said there wasn't enough evidence for a criminal investigation.

Roblox says Simon is not a "credible source of information about our strong safety record."

Another user adopted a new tactic: telling Sega that Sonic was being used by a pedophile on Roblox. The 22 year-old Japanese player wrote to Sega executives and tweeted at the company in public, and soon afterwards Sega filed a copyright infringement notice with Roblox. Sonic Eclipse disappeared in late 2021.

Then, in May 2022 a missing-person report was filed for a 15-year-old girl from Indiana. She'd taken personal possessions and left the state despite lacking the financial means to do so. The detective assigned to the case soon found out from her sister that the girl had been talking to a man she'd met on Roblox going by the name Jacob Shedletsky. Shedletsky had bought one of the girl's drawings, then began sending her gifts and food deliveries.

It took eight days for the authorities to track the girl to Paterson, New Jersey, where they soon found her with Arnold Castillo. Interviewed after his arrest he admitted to using the identities Jacob Shedletsky, Jadon Shedletsky and DoctorRofatnik, and paying a driver $1,000 to bring the girl from Indiana. He admitted to having sex multiple times with the girl over the eight days since her disappearance.

As the FBI investigated how this happened, it was surprised to discover that Roblox users had already exposed Castillo 18 months prior. "Seeing what those children did, as far as compiling all this and their ability to identify Mr. Castillo," says FBI special agent Len Rothermich, "well, they might want to submit some applications to the FBI one day."

In August 2023, Castillo pleaded guilty to transporting a minor across state lines to engage in sex. The court heard that Castillo was a shut-in with "zero social confidence" who shared a bed with his mother, but was talented with computers and worked out he could "make good money" designing Roblox games.

Federal prosecutor Tiffany Preston outlined what she called "every parent's worst nightmare", the eight days in which Castillo's victim was held in a room he'd rented next to his apartment, and repeatedly sexually assaulted. In a victim statement, the girl's sister said she now suffers from depression, anxiety, finds it hard to even leave her room, and refuses to attend school. "These scars will remain with her forever," she said.

The judge sentenced Castillo to 15 years. The scandal prompted a policy overhaul at Roblox, as well as the creation of many new roles in child safety alongside a child exploitation moderation team.

Part of the problem is the scale of policing a platform that has 12 million games and 78 million daily users. Roblox makes a big deal of how it prioritises safety, but one moderator said her team receives "hundreds" of child safety reports daily and simply can't work through them fast enough. Eight current and former trust and safety workers are spoken to in the report, and emphasise inadequate staffing, failure to implement features they recommended (like pop-up safety notices), and the inability of AI moderation systems to discern signs of grooming.

A Roblox spokesperson disputed these claims, pointing to a "robust pipeline" of new safety features. "Implying that a lack of immediate integration of specific ideas, tools or features is a reflection of not caring or lack of prioritization is simply wrong," said the spokesperson.

The real worry for any parent is how many more Castillos are out there. Bloomberg offers data showing that, since 2018, "police in the US have arrested at least two dozen people accused of abducting or abusing victims they’d met or groomed using Roblox." These included "a sheriff’s deputy, a third-grade teacher and a nurse." Specifics include "a man in Florida accused of trying to kidnap a teen he played with on Roblox; a man charged with abducting an 11-year-old New Jersey girl he met on the platform; and a California man who allegedly abused a kid he, too, had met on Roblox."

Roblox said in a statement that the Bloomberg report is awash with "glaring mischaracterisations about how [it] protects users of all ages" and "fails to reflect both the complexities of online child safety and the realities of the overwhelmingly positive experiences that tens of millions of people of all ages have on Roblox every day." Do judge for yourself.

In the court ruling, District Judge Amit Mehta said: "Google has violated Section 2 of the Sherman Act by maintaining its monopoly in two product markets in the United States—general search services and general text advertising—through its exclusive distribution agreements."

Those "exclusive distribution agreements" are deals such as those Google struck to be selected as the default search engine for various platforms, including through Apple's Safari browser, Firefox, Android phones, and some US carriers.

"The default is extremely valuable real estate," Mehta's ruling states.

One of the plaintiff's experts, Dr Whinston, found that "50% of all queries in the United States are run through the default search access points covered by the challenged distribution agreements." Google did not dispute this claim.

In a lengthy 286-page ruling, judge Mehta lays out that not only does Google have a monopoly, it looked to engage in practices to exclude or maintain its monopoly, i.e. through exclusivity deals and uncompetitive ad pricing. It also holds Google was unable to offer justifications for said exclusivity deals.

"After having carefully considered and weighed the witness testimony and evidence, the court reaches the following conclusion: Google is a monopolist, and it has acted as one to maintain its monopoly. It has violated Section 2 of the Sherman Act.

"Google's distribution agreements are exclusive and have anticompetitive effects," it continues. It also states it uses that monopoly to charge "supracompetitive prices for general search text ads", i.e. those above sustainable market prices. It says Google was able to "increase text ad prices without any meaningful competitive constraint."

It's not all bad for Google. The court also found that the company did not have a monopoly power in search advertising. It also said of the search engine giant that it is "widely recognized as the best GSE available in the United States."

A silver lining, of sorts? That was, after all, the pitch of Google's defence. It argued that it was the better product to other search engines.

One quote cited in the ruling is from Apple’s Senior Vice President of Services, Eddy Cue, who reportedly said: "[T]here’s no price that Microsoft could ever offer [Apple]" to switch to Bing. 

Google also managed to avoid a slap on the wrist for its decision to destroy many internal chat messages.

"Any company that puts the onus on its employees to identify and preserve relevant evidence does so at its own peril. Google avoided sanctions in this case. It may not be so lucky in the next one," Mehta says.

That won't do much to stop Google from fighting the ruling. Google's president of global affairs, Kent Walker, said in a statement (via The Guardian): "This decision recognizes that Google offers the best search engine, but concludes that we shouldn’t be allowed to make it easily available."

The US attorney general Merrick Garland called the ruling "a historic win for the American people". The White House also applauded the decision.

We don't know what the repercussions may look like for Google. The ruling does not state any sort of fine or action. However, Google's business cannot continue as is without sanction in light of the ruling, and that could be a big deal for the business of the internet. That's a topic I've been keen to discuss in relation to AI, which only has a light showing in this case, though is being used extensively by Google and others and could drive further inequalities in how search functions. 

The astonishing, but perhaps not surprising, scope of the data scraping was reported by 404 Media, who investigated the documents. It discovered that an internal project codenamed Cosmos (the same name but different to Nvidia's Cosmos Deep Learning service) had staff use dozens of virtual PCs on Amazon Web Service (AWS) to download so many videos per day that Nvidia accumulated over 30 million URLs in the space of one month.

Copyright laws and usage rights were repeatedly discussed by the employees, who found some creative ways to prevent any direct violation of them. For example, Nvidia employed the use of Google's cloud service to download the YouTube-8M dataset, as directly downloading the videos isn't permitted by the terms of service. 

In a leaked Slack channel discussion, one person remarked that "we cleared the download with Google/YouTube ahead of time and dangled as a carrot that we were going to do so using Google Cloud. After all, usually, for 8 million videos, they would get lots of ad impressions, revenue they lose out on when downloading for training, so they should get some money out of it."

404 Media asked Nvidia to comment on the legal and ethical aspects of using copyrighted material for AI training and the company replied that it was in "in full compliance with the letter and the spirit of copyright law."

With some datasets, their use is only permitted for academic purposes and although Nvidia does conduct a considerable amount of research (internally and with other institutions), the leaked materials clearly show that this data scraping was intended for commercial purposes.

Nvidia isn't the only firm to be doing this, of course—OpenAI and Runway have both been accused of knowingly using copyrighted and protected material to train their AI models. Interestingly, one source of video content that you'd think Nvidia would have no problem using is gameplay footage from its GeForce Now service—but the leaked documents show that's not the case.

A senior research scientist at Nvidia explained why to other employees: "We don't yet have statistics or video files yet, because the infras is not yet set up to capture lots of live game videos & actions. There're both engineering & regulatory hurdles to hop through."

AI models have to be trained on billions of data points and there's no way around this. Some datasets have very clear rules for their use, whereas others have fairly loose restrictions, but when it comes to laws on the use of copyrighted materials, it's very clear what can and can't be done, even if the application of it to AI training isn't 100% transparent.

It's not just about copyright, either, as video content often contains personal data. While there isn't a single, overriding federal law in the US that is directly applicable here, there are plenty of regulations concerning collecting and using personal data. In the EU, the General Data Protection Regulation (GPDR) is a law that is expressly clear on how such data can be used, even outside of the EU.

One might also wonder what would happen if a company such as Nvidia is found to have breached various regulations whilst training its AI models—if that system is being used across the globe, would it then be blocked in specific countries? Would the likes of Nvidia be willing to make a new model, trained with all permissions granted, just for those locations? Is it even possible to 'untrain' a system and start afresh with legally compliant data?

Whatever one feels about AI, it's clear that there needs to be a more urgent push for transparency, especially when it concerns the use of copyrighted and personal data for commercial purposes. Because if tech companies aren't held accountable, then data scraping will continue ad hoc.