For many watching Lindsay Clancy’s murder trial, the primary takeaway has been how the deaths of her three small children—Cora, 5; Dawson, 3; and Callen, eight months—were tragically preventable.

Lawyers for the Massachusetts nurse are arguing that she’s not criminally responsible for strangling her children, because she suffered from severe postpartum psychosis, a rare mental health condition that causes new mothers to experience delusions and intrusive thoughts. Defense attorney Kevin Reddington has put Lindsay’s health care providers on the stand, in an attempt to demonstrate how his client repeatedly sought help for her terrifying thoughts, only for her to be minimized, dismissed, and overmedicated at every turn.

For others, including a thriving community of TikTok armchair detectives and conspiracy theorists, the Clancy trial has resulted in a much different takeaway: Lindsay’s husband, Patrick Clancy, was responsible, either literally or by virtue of being a neglectful husband. Many TikTokkers are convinced that he actually committed the murders and framed his mentally ill wife so she would take the fall.

To be clear, there is no evidence whatsoever to suggest this is true. Patrick is not on trial for the deaths of his children, and there has been no indication that he ever will be, or that he is anything other than a deeply grieving father. (Clancy did not respond to a request for comment.) But it’s hard to overstate just how popular this conspiracy theory is, and how rapidly it has overtaken discussion of the trial. The tabloid media has obsessively focused on Patrick, constantly publishing headlines about his life with his new wife Rachel Danis in New York (Patrick remarried earlier this year), and there are countless Reddit threads parsing their social media posts and analyzing his testimony.

Much of the TikTok commentariat’s focus has been on Patrick’s behavior prior to the murders, especially as it pertained to his wife’s postpartum mental health struggles. Per his own testimony, he was not always present at home, even at the height of her suffering. He went to brunch with his friends amidst her struggles with suicidal ideation, and took his daughter on a skiing trip. A month prior to the killings, Lindsay had disclosed to him that she had thoughts of harming the children, but according to his own testimony he had waved them off after she’d assuaged his concerns.

Some experts following the case say that while they understand the impulse to cast suspicion on Patrick, they caution against speculation or cruelly attacking the grieving father of three brutally slain children.

“It’s very easy to try to vilify the husband,” says Leslie Dobson, a forensic psychologist who covers the case on her social media platforms. “But the idea that he actually set her up, or drugged her, or actually killed the children—that’s very extreme.”

But a major driving force behind the antipathy toward Patrick appears to be that women recognize the broad contours of his alleged behavior toward his wife—and they are pissed as hell about it. This is evident when you look at the social media comments on the TikToks accusing Patrick, which are filled with harrowing stories from women whose husbands treated them insensitively while they were struggling with postpartum anxiety or depression, left them for another woman while they were at their lowest, or even emotionally or physically abused them.

(As many TikTokkers have pointed out, Rusty Yates, the husband of Andrea Yates, the woman suffering from severe postpartum psychosis who drowned her children in 2001, reportedly urged her to get pregnant again while she was in recovery from postpartum depression after her fourth pregnancy, against her psychiatrist’s advisement.)

In a nutshell: Cutting-edge hardware and advanced game engines have allowed developers to create increasingly realistic 3D worlds. Even the latest and greatest titles still have a long way to go before virtual environments become indistinguishable from reality, but at the current rate of progress, that eventuality may not be too far off. On the opposite end of the spectrum – and to some, even more impressive – are projects that do more with less, like this walkable city made entirely of ASCII characters.

Grow Now Games, a self-professed one person indie game developer, recently shared the project on YouTube. The dev said his goal was to build a cyberpunk-style city that felt alive, and to build the whole thing using only ASCII characters. So, that’s what he did.

The project uses a small custom engine made using JavaScript and Canvas. The virtual environment is populated with buildings, roads, trees, cars, and pedestrians, and uses camera tricks to work out perspective, depth, and collision for added realism. For something built using only letters, numbers, and symbols, it’s incredibly immersive.

Related: Watch this typewriter pound out ASCII selfies

Games like Minecraft have proven that you don’t need advanced graphics to create a fun and compelling experience, and this is yet another example of that.

Modern ASCII art dates back to the mid-60s but was more widely adopted in the 70s and 80s. Usage declined as computer graphics evolved, but the style remained popular among certain segments of computing enthusiasts and in the hacker community.

The dev said the walkable city is still a work in progress, and noted that he plans to keep advancing the atmosphere, the amount of detail, and interactive elements further. The video already has north of a million views and has generated over 9,000 comments. With any luck, the project will continue and who knows, perhaps it could evolve into the next big thing.

Found is a TechSpot feature where we share clever, funny or otherwise interesting stuff from around the web.

• Google is rolling out a new Gemini for Home update to fix voice commands for Spotify
• The company is also fixing message broadcasting and smart light controls
• Despite the upgrades, users are criticizing Google for not fixing the issues they’ve been experiencing for a while

The transition to Gemini has been rough — even Google knows that, but the latest Google Home update aims to tackle a handful of pain points with voice commands.

In an announcement issued yesterday (August 20), Google shared three fixes coming to the Gemini for Home voice assistant, which is still in early access. Additionally, features in the Google Home app are also tipped to get a glow-up.

Nvidia has poured more than $63bn in investments into other companies.

Nvidia and South Korean chipmaker Rebellions are in talks for a partnership that could result in an investment or an acquisition, Bloomberg News has reported. Talks are in early stages and may not result in a deal, sources told the publication.

According to the publication, Nvidia CEO Jensen Huang and Rebellions co-founder and CEO Sunghyun Park met in person to discuss a potential deal. This comes a year after the company’s chief financial officer Sungkyue Shin said that Rebellions’ “master plan” was to go public.

It raised $400m in a funding round in March, taking it to a valuation of around $2.3bn. Rebellions, at the time, said it is prepping to go public, but did not share details.

Founded in 2020, Rebellions is an Nvidia competitor best known for chips that focus on AI inferencing, as opposed to training. As IBM puts it, AI inference is the ability of trained AI models to recognise patterns and draw conclusions from data it hasn’t seen before.

In 2024, the start-up merged with SK Telecom’s AI chip start-up Sapeon. The idea was that they would together take on global AI giants such as Nvidia.

Nvidia has poured money into dozens of companies over recent years, including frontier developers and infrastructure providers, all in a bid to continue keeping itself central to the AI boom.

In 2025, the company signed US AI company Groq in a licensing deal that saw much of its talent hired away by the chipmaking giant. The start-up announced its first European data centre in Helsinki last year and announced a new $350m funding round earlier this month – though its valuation dropped by nearly half.

Meanwhile, in recent days, the $5.2trn Nvidia has agreed to pay Poolside AI $6bn to license its models and poured an undisclosed amount into Atomic Canyon. Its $63bn investment portfolio, however, majorly consists of Intel and SpaceX, which together amount to more than $50bn.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

After a successful Spectrum Tour, FUJIFILM India has decided to upgrade its Fujifilm X-Space in New Delhi into a three-floor experience center. The new space brings cameras, lenses, Instax, binoculars, QuickSnap cameras, and printing solutions under one roof. It also gives visitors a chance to try products, attend demonstrations, and learn more about Fujifilm’s imaging technology. The center also features Delhi’s first Fujifilm customer service center.

A Three-Floor Center for Imaging Experiences

The first floor of the upgraded Fujifilm X-Space serves as the main experience center, with separate areas for different imaging products. Visitors can try the X Series and GFX System cameras and lenses, along with the complete insta range, Fujinon binoculars, QuickSnap cameras, and printing solutions. The center will also host product demonstrations, workshops, and learning sessions for photography enthusiasts with varying levels of experience.

Moving to the second floor, visitors will find a photo exhibition featuring images taken with Fujifilm cameras. The dedicated studio will host live demonstrations, studio tests, and creative sessions. It will also support demonstrations of film grading and image editing. The third floor is dedicated to customer service, marking Fujifilm’s first such center in Delhi. Users can access product support in the same place where they explore Fujifilm’s imaging products.

Fujifilm X-Space Brings Products and Support Under One Roof

In the upgraded version of the X-Space, the emphasis is on enabling visitors to interact with the Fujifilm imaging technology in an informative manner. The products will not only be viewed but also understood, and their possibilities will be explored. Fujifilm also intends to use the space to promote photography education.

Mr. Haruto Iwata, Managing Director, FUJIFILM Asia Pacific, said, “India is an important market for Fujifilm, supported by a vibrant community of photography enthusiasts and professionals. The upgraded Fujifilm X-Space reflects our continued commitment to bringing our imaging technologies closer to people and creating opportunities for them to explore the possibilities of visual expression.”

Patrice Bouexel explores how AI and automation are impacting employee expectations and behaviours in the wider financial ecosystem.

Now more than ever, we are seeing a landscape in which organisations globally are racing to automate their financial operations. For Patrice Bouexel, general manager for Europe at financial software platform Sis ID, this is evidence of an ecosystem in which companies are struggling to meet expectations.

Bouexel told SiliconRepublic.com, “Finance teams are under the same pressure as every other function – do more with the same headcount and do it faster. 

“Gartner found that AI adoption in finance rose from 37pc in 2023 to 58pc in 2024, while Deloitte’s global survey of more than 1,300 finance leaders found that adopting new technology, including AI, was a top priority for 48pc of respondents.”

He added: “That is not a small shift. It reflects a recognition that manual, paper-based processes cannot keep pace with the volume and complexity of modern payment flows, especially as organisations operate across more currencies, more vendors and more jurisdictions than they did five years ago.” 

Bouexel explained that for the vast majority of companies, automation is no longer a nice-to-have feature that boosts productivity – it is fast becoming a baseline expectation among boards and investors, who often view it as a central way a finance function manages not only risk, but cost as well. 

Rough regulation

But it isn’t solely the introduction of new technologies creating opportunity and chaos for organisations. Among the more pressing challenges for modern-day finance professionals is ensuring that the data utilised and generated by advanced tech is of a high quality and appropriately governed.  

“The gap sits between deploying a tool and trusting what it produces. Organisations that get this right treat automation as a governance exercise first,” said Bouexel.  

“They map which processes are rules-based and genuinely ready for automation, such as invoice processing and expense approvals, and separate those from multi-stakeholder processes that need workflow redesign rather than a bot bolted onto an old process. They also invest in the underlying data before they invest in the algorithm.”

Skills also present a significant challenge, he finds, as AI, automation, data analysis and technology integration are becoming critical to experts in this field.

“Automation succeeds when finance teams are given the time and training to own the exception cases, not just the software licence.”

Sonic speed

He finds, too, that in an environment in which AI and automation are only ever as reliable as the financial data they are relying on, the hastiness to complete a task can often spell disaster. 

Bouexel noted, “An automated process will execute a bad instruction just as efficiently as a good one. If a vendor record is outdated, if bank details were changed without proper verification or if a duplicate invoice slipped through months ago, automation does not catch that. It accelerates it.

“That is the uncomfortable part of the automation conversation that gets skipped over. Organisations are moving faster towards straight-through processing at exactly the point when the data feeding those processes – vendor master files, payment histories and approval trails – is often the least scrutinised part of the finance stack.

“Speed without verified data does not reduce risk. It just means an error or a fraudulent instruction reaches payment faster than it used to.”

He suggested that nowadays, the real goal is prevention rather than cleaning up a mess once it has already happened. As the detection of fraudulent or suspect behaviour becomes a more difficult challenge, largely due to AI and deepfake technologies, verification skills and processes can make all the difference. 

“The more durable answer sits earlier in the chain, verifying who you are actually paying and confirming their identity through a channel the fraudster cannot touch before the payment is authorised, rather than trying to spot the anomaly once it is already in flight,” he said. 

Laudable leadership

When asked the question, ‘What should finance leaders be doing now as fraud, AI and regulation reshape payment security?’, Bouexel remarked that there are three areas for leaders to focus on. 

Firstly, they should separate process risk from technology risk.

Secondly, AI literacy must be built into finance teams as a means of developing a human-based tech skillset, not just into the tools being used. 

He said, “Finance teams need to understand how AI-enabled fraud actually behaves if they are going to recognise and challenge convincing fraudulent requests.”

Lastly, they need to treat regulatory change as a floor, not a ceiling.

“Rules tend to formalise what good practice already looks like. Organisations that wait for the mandate before acting are, by definition, behind.”

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

My colleague Paresh Dave and I each spent a day in federal court in Oakland, California, this week, monitoring the latest Meta child safety trial.

In case you missed it in the spring, Meta (along with YouTube) lost a landmark social media case in California, in which a jury found the companies liable for harming a young user with certain design features in their apps. Then, earlier this month, Meta was ordered to pay more than $940 million in the state of New Mexico for being a public nuisance and causing psychological harm to children.

Now Meta is defending itself in a federal civil trial taking place in the US District Court for the Northern District of California. The plaintiffs in this case are 29 state attorneys general. They’re claiming that Meta violated a federal privacy law protecting children, known as COPPA, by improperly collecting personal information about children under 13 without parental consent. The lead AGs, which span four states—California, Colorado, Kentucky, and New Jersey—also say that Meta made deceptive statements about its platform that were likely to mislead consumers.

Two specific Meta apps are in focus in this trial: Facebook and Instagram.

Meta maintains that it satisfied COPPA law requirements in those states, and denies that its statements about how its apps work were deceptive or misleading. It also insists it is protected from liability by Section 230 of the Communications Decency Act of 1996, which essentially says that platforms like Meta aren’t responsible for the content posted on their apps by users.

I was in the packed courtroom Tuesday to hear opening arguments from both sides, presided over by chief district judge Yvonne Gonzalez Rogers. Per usual, Meta was represented by an army of well-heeled lawyers, who quite ironically kept experiencing technical difficulties with the mic. Megan O’Neill, the deputy attorney general at the California Department of Justice, kicked off the opening statements, laying out how the state AGs believe Meta deliberately hooked kids on its apps and harvested their data.

O’Neill emphasized that throughout the trial, the jury would be shown just “how many features work both individually and together to keep people in the apps. To draw them in and keep them in for longer. Meta has used these features to draw kids onto the apps, and to keep them coming back.”

Meta lead attorney Paul Schmidt then laid out the dozens of safety features that have been built into Meta’s apps over the past several years, and he said the company is committed to improving its apps. He also shifted some responsibility onto users, saying there’s no dispute that some kids find their way onto Meta’s apps and that some teens “struggle to manage their time.” And inevitably some people will post negative content on social media apps, Schmidt said.

Later on Tuesday and then again on Wednesday, the courtroom heard from Arturo Bejar, a former Meta employee and key witness in earlier trials. His main point: Mark Zuckerberg didn’t prioritize youth safety. It was part of company culture during Bejar’s time there that if you had an inkling of an idea for a feature that would spur growth, “you could just test it,” he said. At the same time, he described a culture in which it was “near impossible” to screen products and features for potential harms.

(On Thursday, the trial was put on pause due to a sick juror.)

Meta has been embroiled in a remarkable number of scandals over the past decade, many of which have faded from public memory, been chalked up to the cost of doing business, or were settled out of court. The company’s platforms have played a role in stoking violence and swaying elections. It has regularly siphoned personal data from users in purposefully opaque ways. Not to mention the amount of internal workplace dysfunction at Meta that’s been revealed through depositions, books, and news reports like these.

At the 2026 Chengdu Auto Show, SAIC’s MG brand officially launched its new mid-size sedan, the MG 07, and the pricing might genuinely surprise you. According to CarNewsChina, the car is available in seven configurations, with prices ranging from 108,900 yuan to 159,900 yuan, or roughly $16,000 to $23,500, making it one of the most affordable electric cars.

A surprisingly premium design for a $16,000 sedan

The MG 07 stands out with a star-ring light group up front and semi-hidden door handles for a cleaner silhouette. It measures 4,886mm long and rides on a 2,825mm wheelbase, paired with frameless doors, two-tone wheels, and yellow brake calipers. At the back, there’s a through-type taillight and a 5-stage adjustable electric spoiler. You get six paint options, including Morello Purple, Notting Grey, Oxford Blue, Milia Pink, California Blue, and Salt Lake White.

On the powertrain front, MG is offering both EV and PHEV options. The EV runs on an 800V platform, offering up to 315 horsepower and a maximum range of 845km, while the PHEV pairs a 1.5L engine with an electric motor for a combined range that stretches to 1,745km. The 610km range EV trims interestingly use semi-solid batteries from Qingtao, while every other version sticks with CATL‘s more familiar LFP cells.

The cabin is where things get really interesting

The cabin is really well done. You get a wraparound cockpit with a 15.6-inch floating touchscreen running on the 8295P chip. The driver’s seat comes with heating and ventilation, and the passenger side even gets a “Queen” zero-gravity mode. Add in a multi-functional vanity mirror, a 21-speaker sound system, and mCDC active suspension, and the comfort package feels genuinely thought out. MG has also added LiDAR here, powered by Momenta’s R7 world model.

The MG 07’s price might be its biggest selling point

If you already had your eye on the EV version, pre-sales kicked off last month starting at 125,900 yuan, or about $18,600. Deliveries are staggered depending on which range you pick. Go for the 650km version, and you should have it by the end of August. Opt for the longer 845km range instead, and you will need to wait until October.

Cutting corners: Building a memory chip from the absolute ground up takes years and costs a fortune. ChangXin Memory Technologies (CXMT) apparently decided to skip the hard part – and today, the company stands as China’s biggest DRAM producer. A South Korean court recently heard some fairly wild details about how the memory maker actually got its start.

That origin story came directly from a former Samsung Electronics engineer surnamed Jeon, according to Maeil Business. The veteran employee spent almost three decades at Samsung before jumping to the rival firm around 2016. Jeon appeared in a Seoul courtroom this past week to testify under oath about those early days, telling the judge that his new employer lacked basic laboratory facilities and never planned to do its own independent research when it was founded.

Instead, startup executives reportedly wanted to grab a highly confidential blueprint from Samsung right from day one. Engineers in the semiconductor industry call that document a Process Recipe Plan: it lists roughly 600 specific manufacturing steps, telling engineers exactly what temperatures and equipment pressures to use on the factory floor so they don’t have to guess.

Samsung spent five years and well over a billion dollars dialing in the specifications for its 18-nanometer memory chips. Prosecutors now say the Chinese firm simply recruited a bunch of former Samsung workers to swipe all of it.

By 2023, the company had begun mass-producing advanced memory chips. Today, its output has grown from 40,000 to 720,000 wafers in just a few years, and it’s closing in on Micron.

Unsurprisingly, South Korean authorities haven’t let this slide. Prosecutors indicted ten people late last year for illegally leaking protected national technologies, and a judge handed Jeon a seven-year prison sentence this past April for his role.

Proving all this in court could give Samsung a powerful weapon moving forward. Legal experts believe the company might take sworn statements straight to the US International Trade Commission, a body with the authority to block foreign products from entering the country entirely.