IT General

VisionQuest reunites viewers with Vision (Paul Bettany), the synthezoid first introduced in Avengers: Age of Ultron. However, he's not the only AI from the MCU to reappear in Marvel's latest television outing.

SEE ALSO: We watched 'VisionQuest's first episode at New York Comic Con. It's a trip.

That's because VisionQuest boasts a supporting troupe of Stark Industries AI who have popped up throughout the MCU. Within Vision's simulated mind palace (or mind mansion, really), they manifest as human avatars who help him with everything from basic upkeep to human socialization. From smaller-name robots like Dum-E (Henry Lewis) and U (Jonathan Sayer) to full-on Avengers villains like Ultron (James Spader), here's your guide to the AI characters of VisionQuest.

D.U.M.-E and U

First introduced: Iron Man

Created together during Tony Stark's (Robert Downey Jr.) time at MIT, Dum-E and U are a package deal. They're a pair of hydraulic arms that helped Tony assemble his Iron Man armors.

Within Vision's simulation, the pair manifest as handymen, although VisionQuest lists their more official roles onscreen in a series of helpful bullet points. Thanks to these, we learn that Dum-E is a general maintenance AI in charge of limb servo control and fire suppression. Like Dum-E, U is a general maintenance AI, as well as a defragger and video analyzer.

J.A.R.V.I.S.

First introduced: Iron Man

Arguably the most recognizable Stark AI, J.A.R.V.I.S. (Just A Rather Very Intelligent System) served as Tony Stark's AI assistant and was also uploaded into all the Iron Man suits. Ultron almost destroyed him in Avengers: Age of Ultron, but Tony uploaded what was left of J.A.R.V.I.S. into a synthetic vibranium body, helping create the Vision we know today. Before playing Vision in his physical form, Bettany lent his voice to J.A.R.V.I.S.

In VisionQuest, James D'Arcy takes over the role of J.A.R.V.I.S. His human avatar is a butler-like figure, which is fitting, given that Tony named him after the Stark butler. On top of housekeeping, J.A.R.V.I.S. is a security AI who also takes care of encryption.

SEE ALSO: New York Comic Con 2026: Everything announced at Marvel TV's panel F.R.I.D.A.Y.

First introduced: Avengers: Age of Ultron

With J.A.R.V.I.S. uploaded into Vision, Tony needed a new AI assistant for his Iron Man armor. Enter F.R.I.D.A.Y. (Female Replacement Intelligent Digital Assistant), who took over J.A.R.V.I.S.'s responsibilities from Avengers: Age of Ultron all the way up to Tony's death in Avengers: Endgame.

Originally voiced by Kerry Condon, F.R.I.D.A.Y. is played by Orla Brady in VisionQuest. The show presents her as a socialization AI, which basically means she's Vision's human interaction coach. She also aids him with humor and tactical ops.

E.D.I.T.H.

First introduced: Spider-Man: Far From Home

After his death, Tony left an augmented reality security and defense system named E.D.I.T.H. (Even Dead, I'm The Hero) to Peter Parker (Tom Holland). E.D.I.T.H. (voiced by Dawn Michelle King) lived in a pair of sunglasses that Peter used, then mistakenly gifted to Quentin Beck (Jake Gyllenhaal), aka Mysterio. Following Far From Home, E.D.I.T.H. has been offline.

VisionQuest's E.D.I.T.H. (Emily Hampshire) presents as a major daredevil. As a recreational AI, she tries to understand all the extreme ways humans have fun, leading to some pretty wild stunts. According to the show's bullet-point list of her roles, she also specializes in game theory and weapons.

Ultron

First introduced: Avengers: Age of Ultron

The biggest, baddest AI in the MCU is none other than Ultron. Tony and Bruce Banner (Mark Ruffalo) originally cooked him up as a peacekeeping program, which they then combined with the Mind Stone from Loki's scepter. Following his birth, Ultron decided that the greatest threat to the world was humanity, and set in motion a plan to extinguish all of mankind.

Years after the Avengers defeated him in Age of Ultron, Ultron returns as a prisoner inside Vision's mind, albeit now in human form. Unlike the other AI characters, he doesn't appear to have a specialized role. As the show describes him, he's simply an "asshole."

VisionQuest is now streaming on Disney+, with new episodes Wednesdays.

It is easy to look at a small plastic drive on a desk and assume someone sat down, drew a blueprint, and built it out of thin air. Plenty of devices were genius inventions that one person could lay claim to, sometimes building off the foundation of others. However, hardware history rarely works like that, and combining parts from separate inventions usually turns into a brawl over who crossed the finish line first. The USB flash drive is no different.

SSDs are more than just flash memory chips soldered to a PCB. They’re still relatively simple designs, but they can’t function without a small chip next to the flash that manages how and where data is written, moved, and erased, along with a plethora of other responsibilities.

Generative AI is enabling a tidal wave of nonconsensual intimate imagery and child sexual abuse material. Social media algorithms are exacerbating online harassment and the doxxing of rape survivors. Now, humanoid robots are making the offline world more unsafe for girls and women.

Two teen boys are under investigation by German authorities after allegations that they used a walking bot to grope and harass a 16-year-old girl while she was in public. Police in Cologne confiscated the robot today (Oct. 13) after receiving reports over the weekend. Emergency services were also dispatched to the scene, the Associated Press reported.

The authorities haven't yet charged the boys with a crime, but told the press that they are being treated as suspects.

SEE ALSO: Did Ukraine’s strikes on data centers really silence Russian bots?

The alleged perpetrators were two teens themselves, aged 16 and 19. Witnesses said they appeared to be controlling the 4-foot-tall robot, decorated in human clothing and jewelry, nearby. In video clips that purport to show the two teens, the robot is being steered through busy streets to intentionally bump into women as part of what appears to be a livestream. The robot's behavior is routinely played for laughs in the clips.

This Tweet is currently unavailable. It might be loading or has been removed. This Tweet is currently unavailable. It might be loading or has been removed.

Investigators are evaluating whether or not the robot was programmed to act in such a manner, or if the technology was maneuvered in real time. The answer to that question may determine what kind of legal repercussions the teens face in what is considered uncharted legal waters. Cologne authorities told the Associated Press they were not aware of humanoid robots ever being used for public harassment of this kind.

Popular live streamers have come under fire in recent years for profiting off of the public harassment of strangers, service workers, and women, in particular. Many have used novel technology, such as Meta's AI-powered Ray-Ban glasses, to secretly record interactions for views. It's only recently that platforms have begun to crack down on the viral trend.

My kids bought mood rings in the summer that change color, supposedly to reflect your current mood. This gave me an idea for building a mood ring of my own for my office that would give me useful information at a glance.

Amazon and Google have largely had the smart home world to themselves, at least for those who aren't interested in a solution like Home Assistant. They've not only built a load of hardware, but they also created whole ecosystems with tens of thousands of supported devices between them. They're not without their flaws, but they generally offer a turnkey smart home experience.

Everyone has found themselves scrolling through all their streaming services for ages, struggling to find something interesting to watch. Sometimes, though, it isn't the streaming service that is the problem. Sometimes we avoid watching good content simply because we aren't willing to take a chance on something new. There are tons to watch if you're willing to stick your toe outside your comfort zone.

SAVE $1,302: As of Oct. 13, the 65-inch TCL QM8L SQD-Mini LED 4K TV is down to $1,197.99 at Amazon. That’s a 52% discount off the $2,499.99 list price on Mashable’s "best QLED TV for most people."

Opens in a new window Credit: TCL TCL 85-inch Class QM8L Series TV $1,197.99 at Amazon
$2,499.99 Save $1,302.00   Get Deal at Amazon

Shopping for a new TV often feels like a dizzying exercise in decoding endless display acronyms. But if you want staggering peak brightness and deep contrast without taking on a multi-thousand-dollar OLED price tag, TCL’s flagship 2026 set is currently discounted. The 65-inch TCL QM8L SQD-Mini LED 4K Smart TV is down to $1,197.99 on Amazon, saving you a massive $1,302 off its original list price of $2,499.99.

In Mashable’s guide to the best 65-inch TVs, shopping reporter Leah Stodart dubbed the TCL QM8L the "best QLED TV for most people." Unlike base-level sets that struggle in low-light environments, the QM8L introduces TCL’s revamped Super Quantum Dot (SQD) backlight system. By using microscopic 5-nanometer particles instead of standard 60-nanometer dots, the display delivers pinpoint color accuracy, dynamic contrast, and anti-blooming technology that actually keeps halo effects at bay during dark movie scenes.

Clocking in at up to 6,000 nits of peak brightness alongside over 4,000 discrete local dimming zones, this TV is tailor-made for high-glare living rooms. Stodart noted that its sublime peak brightness makes it "one of the brightest TVs you can buy," easily handling midday glare and sunlit sports matches while offering contrast that rivals far more expensive non-OLED displays.

The 2026 QM8L is built for high-end entertainment and gaming. It packs a 144Hz refresh rate, AMD FreeSync Premium Pro, and TCL's Game Accelerator suite for smooth console action, while Google TV with Google Gemini AI handles smart navigation.

Grab the 65-inch TCL QM8L SQD-Mini LED 4K Smart TV at Amazon before the 52% discount vanishes.

AI-generated images can now give celebrities entire social media moments they never took part in. For Mia Goth, that meant a supposed response to her divorce — right down to a T-shirt and breakup soundtrack.

After news of the Pearl and MaXXXine actress's divorce filing from Shia LaBeouf a few weeks ago, an Instagram post appeared to show Goth wearing a shirt that read "MY EX IS MY BIGGEST FAN." Paired with Beyoncé’s "Single Ladies (Put a Ring on It)," it was easy to read as a comment on the split.

Except Goth’s representative says the image was AI-generated. And Goth doesn’t even use social media.

The post came from @miagoth_, a verified fan account with more than a million followers. Despite the fan-page label, People initially wrote about the image as a selfie Goth had posted herself, describing the shirt as her fashion choice days after filing for divorce. (The article and post have both been removed.)

SEE ALSO: Taylor Swift's wedding photos are going viral. There's just one problem.

That confusion is exactly what the actress's team says it has been trying to stop.

"Mia Goth does not have ownership of, nor does she engage in social media on any platform," her representative, Christine Tripicchio, told The Verge. "We have filed multiple complaints with Instagram to remove the verified fan account which continues to post counterfeit, AI images of Mia Goth."

Tripicchio also questioned how the account had qualified for a verification badge in the first place.

"The fact that this fan account meets Instagram’s criteria to qualify for a 'blue check' is disconcerting on a fundamental level," she said. "Beyond that, it is incomprehensible that Instagram evidently prioritizes the dissemination of forged and manipulated images over the protection and integrity of a public figure’s likeness."

In response, Meta told Mashable, "This account has been removed for impersonating a public figure."

"We must call out the misuse of AI"

Goth’s team isn’t alone in trying to correct a public moment that never happened.

A March investigation by Indicator identified six AI-generated influencer accounts with roughly a million Instagram followers combined that posted fabricated celebrity encounters. One, @ayannasoblack, shared fake gym photos with Dwayne "The Rock" Johnson and John Cena. All but one of the accounts linked to Fanvue profiles selling access to synthetic adult content.

Real performers' work also supplies material for these accounts. As Mashable reported, AI personas have recently reproduced comedians' jokes and performances to build audiences. Comedian Sarah Purdum told Mashable that women’s material was being "stolen and repurposed for sexy AI models."

SEE ALSO: Fake influencers, real money: Inside the AI creator business

For celebrities, the consequences extend beyond invented encounters. In January 2024, nonconsensual, sexually explicit AI-generated images of Taylor Swift spread online, prompting SAG-AFTRA to condemn them as "upsetting, harmful, and deeply concerning." Later that year, Swift cited a separate fabrication — images falsely suggesting she supported Donald Trump — when announcing her endorsement of Kamala Harris.

"It really conjured up my fears around AI, and the dangers of spreading misinformation," she wrote.

Scarlett Johansson raised a similar objection in 2025 after an AI-generated video depicted her and other Jewish celebrities protesting Ye’s antisemitism. Although she condemned antisemitism, she objected to being included in a video she never participated in. "We must call out the misuse of AI, no matter its messaging, or we risk losing a hold on reality," she said in a statement.

Even deceased performers' families are being left to push back: In October 2025, filmmaker Zelda Williams asked people to stop sending her AI-generated videos of her father, Robin Williams, saying the recreations were "NOT what he’d want."

What is Instagram's current AI policy?

An image being AI-generated doesn’t automatically mean Meta will take it down.

In its explanation of its policy, Meta says it generally keeps manipulated content available unless it breaks another platform rule, such as those against bullying or harassment. Instead, it uses "AI info" labels to provide context when it detects certain technical signals or someone discloses AI use.

SEE ALSO: Meta leans into AI in your feeds

Even those labels have needed adjusting, though. In 2024, Meta changed "Made with AI" to "AI info" after acknowledging that some photos with minor AI retouching were getting flagged. It later moved labels for content detected as AI-edited into the post’s menu, while keeping visible labels on content detected as AI-generated.

But that approach depends on detecting AI use in the first place. Meta has acknowledged that its systems can’t identify every AI-generated image and that the invisible markers it uses for detection can be removed. A missing AI label doesn't necessarily mean a photo is real (clearly).

Back when I had a cheap Wi-Fi router, I bought a Wi-Fi extender to compensate for the router's short range. But there was a problem. Its SSID and password were different from my normal router, so I had to connect manually each time I used the extender. The better solution for this was a mesh Wi-Fi system at a higher cost. At that time, I didn't even know that a mesh network was different from Wi-Fi extenders.

Need to print a document from your Googlebook? Thankfully, the process of connecting to a printer takes less than a minute, and although most Googlebook apps lack any printing functionality, Chrome and other standard Google tools will let you print a page or document with a simple keyboard shortcut.

Raye is a British singer-songwriter who planned on holding an intimate one-night concert for her fans next week.

The event's venue has capacity for 5,000 attendees. There were 1.8 million sign-ups for tickets. It sounds like a good problem to have, right? Having that many fans? One problem. The vast majority of those sign-ups were AI bots, according to Raye, run by profit-seeking resellers hoping to make money off her fans.

"This Symphony May Contain Hope" is a planned one-night live concert by Raye at the Royal Albert Hall in London on Oct. 23. It appears the singer was expecting to receive thousands of sign-ups and then send out a pre-sale link for interested parties, with tickets available on a first-come, first-served basis.

This Tweet is currently unavailable. It might be loading or has been removed.

However, Raye wasn't expecting the resellers and AI bots to hit as hard as they did.

"Sadly our sign-up process was overwhelmed by scalpers, using AI bots to create hundreds of thousands of fake registrations," Raye wrote on her X account. "It's a new world we are living in now, I have never witnessed bots on this scale."

"It is so important to me that fans can be tickets at the face value price," she continued.

Raye says that for this special event only, she and her team would manually review the sign-up list and send a pre-sale link and code to individuals that they believe are real fans.

"I'm not sure what the future holds for ticketing, I do know this experience has really opened my eyes," Raye said.

Resellers have always been a problem when it comes to tickets for concerts, tours, and sporting events. However, the rise of AI has helped resellers streamline and scale their ticket buying, often leaving few, if any, regular-price tickets for actual fans.

Some artists have been forced to get creative in how they sell tickets to ensure availability. Spotify recently released features to help artists identify their real fans and offer them first dibs on tickets to local shows.

In the United States, the Federal Trade Commission recently sued Ticketmaster, which controls a majority of the U.S. concert market, for allegedly working with resellers to inflate ticket prices. Ticketmaster denied the accusation, but a jury later found that the company gouged customers and abused its monopoly power. Still, a long-term solution to the reselling problem is desperately needed.

While scrubbing through AI bot accounts may be possible for a special one-off event, it doesn't seem feasible for months-long nationwide or even global touring events.

Google Maps Timeline is one of those features that can be incredibly useful when needed. It can remind you where you went on a particular day, the places you visited on vacation, and even the routes you took from A to B. I never liked the idea of Google knowing everywhere I ever went, so I replaced Google Maps Timeline with a free, open-source alternative that stores location history on my local hardware.

Setting up a brand-new computer should feel clean and fast, but modern operating systems treat your personal desktop like an advertising billboard from the second you log in. It gets frustrating watching basic hardware stutter under background tasks you never wanted or asked for. Unfortunately, the best way to make Windows 11 faster is to turn off features it tries to force on you.

If you're looking for a hand tool that's tough, durable, and reliable, with a lifetime warranty that won't break the bank, there's a brand at Harbor Freight you don't want to overlook. I'm talking about Icon tools, and the lineup includes several great hand tools you'll absolutely love.

Google's Pixel phones have always been renowned for their excellent cameras. But what frustrates me is that what you see in the viewfinder doesn't always match what you get after you press the shutter button.

Google announced the public launch of Gemini 4 Argon after an earlier limited release to select testing partners. It's the first model in the company’s Gemini 4 generation and now sits at the top of Google’s lineup.

Google CEO Sundar Pichai announced the launch on X, where he said the model would be available globally starting today.

This Tweet is currently unavailable. It might be loading or has been removed.

Google first announced Argon on Sept. 30 in a blog post by Google DeepMind SVP Koray Kavukcuoglu. At that point, it was limited to trusted cyber defenders in Google’s Fairwind Program. With today’s release, Google AI Pro and Ultra subscribers can use Argon in the Gemini app.

Google says Argon is built for long, complex work in software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense. For everyday chat, Google says the model gets to the point with less hedging, repeats itself less, and avoids clichés. On longer tasks, it gives more frequent progress updates.

Argon 4 arrives during a busy fall season for frontier AI models. Google repeatedly missed its own deadlines for the launch of Gemini 3.5 Pro, and meanwhile, Anthropic released Claude Opus 5.5, and OpenAI released GPT-6 Astra. Meta's "super intelligence" Muse Wonder was just released yesterday.

As expected, Google’s own evaluation shows Argon testing ahead of its rivals on many benchmarks. On the independent Artificial Analysis Intelligence Index, Argon ties GPT-6 Astra at 53 with high reasoning enabled, one point above GPT-6.1 Sol and the new Muse Wonder Max. The same index places it about five points behind Claude Opus 5.5 and Claude Sonnet 5.5.

The latest Artificial Analysis Intelligence Index, as of Oct. 13. Credit: Artificial Analysis

Coding is the clearest focus of the release. Google says Argon already handles much of its internal engineering work, and today’s update brings Google Antigravity’s agentic coding interface directly into the Gemini app on macOS.

Want to learn more about getting the best out of your tech? Sign up for Mashable's Top Stories and Deals newsletters today.

Google Gemini Argon 4: How to try it

Starting today, Google AI Pro and Ultra subscribers can use Gemini 4 Argon in the Gemini app. Google AI Pro costs $19.99 a month. At I/O 2026, Google added a $100-a-month Ultra plan and lowered its top Ultra tier from $250 to $200 a month.

Google set an introductory API price for Argon:

• $2 per million input tokens

• $10 per million output tokens

• Cached input tokens at 95 percent off, or $0.10 per million

That is less than both main rivals charge. Anthropic's Claude Opus 5.5 costs $4 and $20 per million input and output tokens, while GPT-6 Astra costs $10 and $50 per million tokens. Today’s announcement focuses on the Gemini app and doesn’t say whether API access has widened beyond Google’s earlier rollout plans.

The rollout comes with several other changes to the Gemini app. First, Antigravity’s agentic coding interface now lives inside the Gemini app on macOS, so users can move between chatting and coding without switching apps. Second, a new toggle in the main chat window switches Gemini from chat to tasks. In tasks mode, Gemini can keep working across your Google apps and other services while you’re away from your computer. Pro and Ultra subscribers get access.

Additionally, Gemini now connects to more than 40 apps, including Instacart, OpenTable, Adobe, and Shopify. Google pitches this for jobs like ordering groceries, booking restaurant tables, and designing or building an online business.

Google Gemini Argon 4: Benchmark performance

Google’s comparison table has 18 results. Argon has the top score on 14, ties with GPT-6 in cybersecurity, and trails on five. All figures in this section come from Google. Typically, AI companies release results that show their own model scoring above its competitors.

Credit: Google / Mashable screenshot

Argon’s biggest leads are in knowledge work. Google says it tops the Vals Index, which weights finance, coding, legal, and tax work by each sector’s share of U.S. GDP. On AutomationBench, Zapier’s test of end-to-end business tasks, Argon scored 51.3 percent. That is about nine points ahead of the next model.

The coding results are mixed. Argon set a new high on DeepSWE v1.1, Google’s main test of long software engineering tasks, at 77.9 percent. But it trails GPT-6 Astra on FrontierSWE v2. On Terminal-Bench 4.0, it trails all three rivals, and Opus 5.5 leads.

Some of Google’s figures for competitors don’t match what Anthropic reported. Mashable reported that Anthropic put GPT-6 Astra at 57.9 percent on Terminal-Bench 4.0, while Google lists 58.2 percent. On the Terminal-Bench Science benchmark, Anthropic reported 58.7 percent for Opus 5.5 (default effort), while Google lists 63.3 percent. When Anthropic launched Opus 5.5, it also cautioned that "benchmark margins have become a less reliable guide to real-world differences."

Independent of Google, the Artificial Analysis Intelligence Index has tested Argon’s "High" setting. For testing, Artificial Analysis combines 10 evaluations covering math, science, coding, agentic tasks, and reasoning into a single composite score. Artificial Analysis says it runs all of the tests itself on standardized hardware, so it isn’t relying on each company’s self-reported numbers.

On that index, Gemini 4 Argon (High) scores 53, well above the median of 26 for comparable models. While still behind heavyweights like Claude and ChatGPT, Argon 4 represents a 23-point jump over Gemini 3.1 Pro Preview, Google’s previous top model.

Argon ranks just a hair above Muse Wonder but below Anthropic and OpenAI's latest. Credit: Artificial Analysis

At introductory pricing, Argon costs about $1.99 per Intelligence Index task, roughly 60 percent of GPT-6 Astra's cost. Once the discount ends, that is expected to rise to $3.98, about 1.2 times Astra’s cost. By comparison, Opus 5.5 costs $5.98 per task.

Google Gemini 4 Argon: Safety and safeguards

Before today’s consumer rollout, Google released Argon in phases, starting with cyber defenders, while it participated in the U.S. government’s voluntary process for pre-release model access.

For everyone else, Google says the model is designed to refuse requests that could aid cyberattacks or chemical, biological, radiological, or nuclear attacks. Google says red teams tested those safeguards.

Google also says Argon is its most resistant model yet to indirect prompt injection. On Gray Swan’s benchmark — an Injection Prompt Test that measures how easily a model can be hijacked by malicious instructions hidden in content it reads — Google reports that attacks succeeded 0.7 percent of the time against Argon. The figures were 1.0 percent for both Opus 5.5 and Fable 5.1 and 8.5 percent for GPT-6 Astra. Lower is better on this test.

You can find the full Gemini Argon 4 model card on the Google website.

Disclosure: Ziff Davis, Mashable’s parent company, in April 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.

On Tuesday, Democratic Sen. Maggie Hassan released the results of her investigation into romance scams on Match Group apps, including Tinder and Hinge.

Hassan, the New Hampshire Democrat who serves as ranking member of the Joint Economic Committee, began this investigation in Sept. 2025, when she and Republican Sen. Marsha Blackburn sent a letter to Match and its CEO, Spencer Rascoff, requesting documents about romance scams on the company's platforms.

SEE ALSO: AI romance scams: Follow these tips to avoid being scammed online

These scams occur when a bad actor gains someone's trust by pretending to be romantically interested in them, only to then ask for money. The Federal Trade Commission reported that Americans lost almost $1.5 billion to romance scams last year.

The investigation — called "Outmatched: Tinder, Scammers, and Match Group's Efforts to Catch Up" — focused on how the company detects and removes scam accounts. As part of it, Match Group provided the committee with two written responses and internal documents in Oct. and Nov. 2025, and additional documents throughout this year. Match Group's senior vice president of trust and safety, Yoel Roth, also sat down for an interview with committee minority staff in April 2026.

The investigation found that, as of the second half of 2025, Tinder hadn't updated most of its AI models used to detect scammers and other bad actors for years. Roth told the committee that Tinder performs ongoing maintenance on its machine learning models and that Match retrains its models every year or every couple of years.

According to data Match Group provided the committee, the number of views received by the median scammer’s profile on Tinder globally increased by more than 700 percent between May 2024 and December 2025. In the first nine months of 2025, almost 1.5 million accounts belonging to scammers or other bad actors on Tinder and Hinge in the U.S. matched with at least one user; nearly 500,000 of those accounts exchanged contact information with another user. The two apps banned tens of millions of inauthentic accounts in 2025.

In Oct. 2025, Tinder announced the U.S. rollout of its user verification process, Face Check, which became mandatory for all new U.S. users in early December. Hinge followed suit. Match Group told the committee that within a month of nationwide deployment, Face Check reduced user interactions with spam, bots, and scammers on Tinder in the U.S. by more than 70 percent. The company said that the reduction was sustained for at least the next seven months.

Face Check isn't perfect. A viral YouTube video from April demonstrated that it can be tricked, and a Tinder spokesperson acknowledged these concerns to Mashable at the time. The spokesperson added that the app is committed to continuously improving its systems.

SEE ALSO: Tinder responds to viral video about tricking facial scan

Match also introduced a new AI model to detect scammers and other bad actors in the first half of this year after continued follow-up from the committee, according to the report.

Roth told Mashable in an emailed statement that Match welcomed recognition of its efforts to protect users, including Face Check, but disputed the report's characterization of its work: "The report mischaracterizes both the historical scale of our investment and the work already completed."

"Alongside our engagement with the Committee over the past year, we've made the largest safety advancements in Match Group's history. Trust and Safety is a foundational priority, and we intend to continue leading the industry in building tools that promote authentic, safe connections," Roth said.
 
Roth also said that romance scams are a cross-platform, societal problem that no company can solve alone, which is why Match participates in and helps fund cross-industry efforts to combat online scams. Those efforts include membership in the Aspen Institute's National Task Force on Fraud and Scam Prevention, co-founding the Tech Against Scams Coalition, serving on the advisory board of the Global Anti-Scam Alliance, and supporting Sen. Hassan's ReportScams.gov Act.

Hassan said in a press release shared with Mashable: "During the course of our investigation, Match Group made progress in combating scammers — but it shouldn't have taken so long, and there's still more that Match needs to do."

You don't need to spend iPad-level money in order to get a highly productive, high-performance tablet that can do everything you could possibly need: stream your favorite app's TV shows, work in WPS Office, take notes, or check your work emails. That's why something like the 11-inch Lenovo Idea Tab is such a fantastic choice, as it packs a sharp 2.5K display, 8GB of built-in RAM, and expandable storage. An even bigger reason to grab one? Its current $90 discount and included stylus pen make this one deal you won't want to pass up.

If you simply can't get enough of Grand Theft Auto VI prior to its Nov. 19 release, there are still new GTA 6 leaks coming out.

Cyberleek, the anonymous entity that has been leaking clips from a pre-release build of GTA 6 since August, made their grand return in October after about a month of silence. Last week, Cyberleek posted a 24-minute video that seemed to show some story sequences and a glimpse of the male protagonist Jason rendered completely nude. That wasn't the end of it.

And, as Cyberleek said they would, nude clips of the female protagonist Lucia soon followed. On top of that, Cyberleek is now threatening to release a development build of the game, and according to Dextero, may have already done so. Websites affiliated with Cyberleek include a link to a file named GTA 6: Development Build v1.0.1, locked behind a $40 paywall.

It should go without saying that downloading files from a sketchy website is not a wise idea, nor is purchasing illegal content that violates Rockstar's copyright.

SEE ALSO: 'GTA 6': Is PS5 Pro the best way to experience the game at launch? GTA 6 Cyberleek update: What has been posted this week?

Cyberleek returned this week with more clips that you definitely should not watch at work.

That's because these new clips all center on the female protagonist, Lucia, nude. In the clips, Lucia does mundane things like walk around Vice City, drive a vehicle, and parachute off a skyscraper while wearing little to no clothes (OK, so not entirely mundane, actually). It doesn't seem like anything new has been posted since Monday morning, though.

One thing to note about these clips: The nudity doesn't appear to be a normal element of the gameplay, but rather some type of mod. Rockstar has designed its last couple of big open-world adventures around extremely reactive NPCs, and none of them react at all to seeing Jason or Lucia nude in these leaked clips. Between that, and another clip from last week where the player spawns a tank out of thin air, it sure seems like developer tools are being used to achieve outcomes that will not be possible in the final game.

In other words, don't expect to be able to streak around Vice City in the nude on Nov. 19.

Cyberleek threatens to release a development build of the game Credit: Rockstar

On websites associated with Cyberleek, the leaker threatened to release a full playthrough of the entire GTA 6 story if their memecoin reached $30 million in market value. The leaker also threatened to release a development build of the game if Rockstar failed to meet their demands to release GTA 6 on a physical disc. There's just one problem: The coin is tanking.

IGN reported a market cap of just $1.7 million, well below any of the stretch goals specified by Cyberleek, and certainly not even close to the $7.8 million high it reached on Oct. 11. Despite this, Cyberleek may have made good on their threat to release a development build of the game.

Dextero is reporting that the development build has been posted online, locked behind a $40 paywall. In addition, a Cyberleek website with all of the leaked clips now includes this update, posted on Oct. 13: "DEVELOPMENT BUILD IS OUT: GTA 6: Development Build v1.0.1.08022025 is up. The deal was 30 million marketcap and the build goes out. Cap was not hit. Build is out anyway. $40 per download. Pay it, then take the mirror."

Because the leaker is anonymous, and because it would be illegal and dangerous to download mystery files from an unverified website, we can't confirm that this development build is real, or that it was even directly posted by Cyberleek.

However, if there is now a development build floating around the web, expect to see far more GTA 6 leaks in the weeks leading up to the Nov. 19 release date.

Alternatively, you could wait and play the game when it actually comes out. You can preorder the game now.

What could Cyberleek release next?

A Cyberleek website includes a poll about what type of Lucia content to release next. The poll is set to end on Oct. 14, with the leading option as "Race track."

It seems extremely likely that more clips featuring Lucia are set to be released any day now.

Want to learn more about getting the best out of your tech? Sign up for Mashable's Top Stories and Deals newsletters today.