In a nutshell: Ferrari’s first electric vehicle sold for $40 million at a charity auction in California, setting a record for a new vehicle sold at auction. The result offers an early sign that Ferrari’s brand power, and its appetite for limited, customized EVs, remain strong despite a contentious launch.

RM Sotheby’s sold the first production example of the Ferrari Luce, a customized version of the company’s new electric model. The buyer was not identified. The proceeds will go to the Ferrari Foundation’s education programs.

The sale price was more than 35 times the Luce’s usual price, and it also topped Ferrari’s previous record for a new vehicle sold at auction. In 2025, a bespoke Daytona SP3 brought in $26 million for Ferrari’s education initiatives.

The Luce is Ferrari’s first move into the electric vehicle market. It was developed with input from former Apple designer Sir Jony Ive and was one of the company’s most closely watched launches in recent years.

The model has drawn criticism since its May debut. Some critics said the five-seat, battery-powered vehicle moves too far from the designs that made Ferrari famous. Italy’s deputy prime minister Matteo Salvini and former Ferrari chairman Luca Cordero di Montezemolo were among those who objected. Montezemolo said Ferrari was “risking the destruction of a legend.”

Ferrari chief design officer Flavio Manzoni said in May that criticism comes with innovation and that he believed people would eventually appreciate the Luce.

The launch received unusual public attention. Italian President Sergio Mattarella and Pope Leo were invited to see the vehicle. Ferrari’s shares, however, fell the day after the unveiling as criticism of the car spread.

Ferrari has not disclosed a sales target for the Luce, though the Financial Times reported that the company has already reached its goal for the year, helped by demand from China.

Sotheby’s said the auction gave one buyer an “unbeatable opportunity” to own the first production Luce. The auction house called the sale a “tangible expression of innovation, responsibility, and long-term vision for future generations.”

Ferraris regularly bring high prices at auction, especially limited-production and customized models, though the $40 million sale falls short of the all-time auction record for any vehicle. That belongs to a 1955 Mercedes-Benz 300 SLR Uhlenhaut Coupé, which sold for $142 million in 2022.

A new analysis of the 60 largest data centers under construction in the US by Amazon, Google, Meta and Microsoft estimates the sites could produce a cumulative 101.5 million tons of carbon dioxide annually once they’re fully up and running, Financial Times reports. That’s about 7% of all 2025 US power-sector emissions. Put another way, it’s the equivalent of running 27 coal power plants or putting 24 million more gas-powered cars on the road.

The backlash against America’s data center and artificial intelligence boom has been building for a while now, reaching a fever pitch over water use, land grabs and noise complaints. Recently, environmental activist Erin Brockovich has been mapping out communities affected by data center construction and operations, and polling shows that most Americans don’t want AI factories anywhere near them.

Now, it appears there’s a looming carbon problem to add to the pile.

Big Tech companies appeared publicly committed to cutting emissions just five years ago, making the new findings feel like a notable about-face. Financial Times found that Amazon’s emissions climbed 16% from 2024 to 2025, driven by data center construction and a 34% jump in electricity purchases. Microsoft reported a 25% increase in estimated emissions during the same period, while Google’s parent company, Alphabet, reported an 18% year-over-year increase in its emissions figure according to its own adjusted metric.

Researchers point to a mix of causes, from the sheer scale of demand for AI infrastructure to a political environment under the Trump administration that dramatically rolled back climate rules and clean energy tax credits in favor of fossil fuels.

On the grid side, the numbers tell a similar story. US gas-fired energy capacity under development nearly tripled in 2025, with FT’s analysts predicting the country’s existing gas-burning infrastructure could grow up to 50% if everything currently planned is built.

Researchers say it’s not that clean energy additions have stalled entirely; rather, utilities are racing to meet data center demand with power that’s available now, and that runs around the clock. Fossil fuels (like gas and coal) fit the bill better than wind or solar until large-scale storage tech catches up.

That’s left the tech giants leaning harder on renewable energy credits to make up the dirty differences between their public commitments and what’s actually happening on the grids. But doing so is expensive and exactly the sort of thing likely to be scaled back over time as profitability pressure mounts.

Utilities, meanwhile, are increasing production to meet the mounting and projected demand, preparing to roll out tens of thousands of megawatts of energy capacity across the US specifically to serve data center infrastructure. Three-quarters of the operators serving these 60 sites are building or planning new gas capacity, and a third of those running coal plants are pushing back retirement dates.

This fresh wave of gas and coal plants occupies a small but rapidly growing portion of the US energy generation mix — trending in the opposite direction to what experts expected only a few years ago. The resulting environmental impacts and grid strains from these energy decisions are stacking up to just one more AI boom cost that nobody outside the tech industry signed up for.

Antuan Goodwin

Senior Writer, Electrified Cars

Antuan has nearly 20 years of expertise and experience testing hundreds of cars, including electric, hybrid, plug-in hybrid, hydrogen, and traditional combustion vehicles.

See full bio

Advertisement

Fireflies.ai is expanding beyond AI meeting notes with a new Email Assistant designed to help users manage their overflowing inboxes. The tool can automatically sort incoming emails, draft replies, and create follow-ups by using context from both email conversations and previous meetings. Here’s everything you need to know about it.

Fireflies Email Assistant Connects Email and Meeting Context

The biggest difference between Fireflies’ Email Assistant and a traditional AI email tool is the context it can use. Instead of relying only on what’s written in an email thread, the assistant can also look at past meetings with the people involved in the conversation.

For example, if a customer follows up about something discussed during a previous meeting, Fireflies can use that meeting context to help draft a response. This means users don’t necessarily have to search through old meeting notes or try to remember what was discussed.

Fireflies has included several features aimed at reducing repetitive inbox work:

• Auto Labels: Automatically sorts and labels incoming emails so users can identify priorities more easily. It does not modify or delete existing emails, workflows, or rules.
• Auto Draft Replies: Creates draft responses using the current email thread and relevant past meetings. Users can also set custom Writing Instructions to control the tone and style. Nothing is sent without user approval.
• Follow-Ups: If an email doesn’t receive a response, the assistant can draft a follow-up in the original thread after a user-configured waiting period.
• Daily Briefs: Brings email insights and action items into the same daily overview as meetings and Slack activity.
• AskFred: Users can ask questions in plain English to find emails, summarize conversations, or retrieve action items from the week. The assistant can then drop a relevant draft into the inbox.

The company says its tools now work across meetings, email, Slack, and CRM systems. The idea is to connect information across these services and automate some of the work that happens after a meeting, such as drafting replies, updating records, and flagging important tasks.

Fireflies Email Assistant is available immediately on all Fireflies plans, including the free plan. Users can connect a Gmail or Outlook account through the Email Connector inside Fireflies to get started. For users who need additional features, Fireflies’ Pro plan starts at ₹955 per seat per month when billed annually, according to the company.

We’re halfway through The Big Bang Theory spinoff Stuart Fails to Save the Universe, and I still maintain the opinion that it’s better than the original CBS sitcom.

But that hasn’t stopped the two from being heavily intertwined. Over the last four episodes, we’ve seen cameo appearances from Raj (Kunal Nayyar), Penny (Kaley Cuoco), Bernadette (Melissa Rauch) and Amy (Mayim Bialik). Could this week add a fifth member to the growing party?

from the one-more-time-into-the-fray dept

A few years ago a California appeals court produced the unfortunate Liapes v. Facebook decision. In this decision, it gave the go-ahead to age discrimination claims under California’s Unruh Act resulting from ads not being served to users of certain ages. It’s a nonsensical application of anti-discrimination law, because ads are not themselves the goods and services that the Unruh Act governs but instead expression. Furthermore, for more than a century American media has been supported by an advertising model that allows advertisers to target audiences based on their demographic information, including their age. It would upend the media ecosystem if doing so can suddenly tempt legal liability, so surely the First Amendment has something to say about applying anti-discrimination law to expressive businesses.

And so should Section 230, because the decision didn’t involve suing the actual advertiser who chose to target certain potential customers, but Facebook itself, even though it was merely acting as a platform intermediating third party-generated content—in this case, the ads that advertisers wanted to carefully direct in a way they thought would get them the most bang for their buck. Intermediating third party-generated content—even advertising content—is a core activity protected by Section 230, and it threatens to upend the Internet ecosystem if platforms are no longer legally able to do it.

But that was then, and back then the Copia Institute submitted an amicus letter urging the California Supreme Court to review the appeals court decision that reached this troubling conclusion. Unfortunately, the Supreme Court did not take up the review, and so the Liapes decision remained on the books, tempting other courts to follow it down its troubling path.

Which a California superior court has now done, this time allowing a similar sort of claim to proceed against Google based on certain users not getting served all the ads that they would have liked to see. Google appealed the decision, and the Copia Institute this week filed an updated amicus submission pointing out the same infirmities with it that we had in the Liapes case, namely that the First Amendment protects the choices speakers make in how to direct their expression to audiences, and, even if it did not, Section 230 still protects platforms handling expression provided by others from anything that might be wrong with that expression, including that it might violate state anti-discrimination law. But the decision by the superior court here, letting these Unruh Act claims to go forward, essentially ignored all of that statutory and constitutional protection that all expressive businesses need, be they traditional media entities or Internet platforms.

This time around we also pointed out yet another reason why that editorial discretion exercised in choosing how to direct speech must be constitutionally protected: because state actors don’t always agree. Some can want speech to be directed one way, and some can want it to be directed in another, and the First Amendment exists to keep expressive entities from getting caught in a tug-of-war between them. Especially because, as this case shows, it may not even be different state entities who disagree; here it is the state of California itself, who on the one hand has one law suggesting it might be illegal to refuse to provide certain speech based on age, but at the same time has been passing a slew of new laws demanding platforms do exactly that: refuse to serve certain expression based on age.

As the Copia Institute has been arguing, both laws are unconstitutional, and this conflict illustrates why the First Amendment must protect the expressive choices about whom to provide what expression to from any such governmental meddling.

Filed Under: 1st amendment, aadc, advertising, california, liapes, section 230, targeted advertising, unruh act

Companies: google

Justin Ross has given Mega Man fans something they have waited years to play again. Mega Man X: Regenesis launched today as a free download, and the finished product lands closer to Capcom’s classic 16-bit entries than most people expected from a solo passion project.

You find yourself in a universe where the timeline has taken a dark turn, beginning shortly after the original Mega Man X. Sigma has done the unthinkable; X has vanished, and humanity is on the verge of destruction. Dr. Cain’s last-ditch effort, Project Regenesis, begins, offering X a second chance to turn the tables and defeat the Mavericks who now control the situation.

My Play Watch Mega Man Retro Gaming Smartwatch, Built from Mega Man™ 2

• Built from Mega Man 2 Authentic sprites, stages, bosses, and music rebuilt for your wrist
• Multiple Game Modes Classic Mode with full stage progression + Dr. Wily’s Castle, Arcade Mode for endless, high-score gameplay plus Play Time…
• One-Touch Action Gameplay Auto-run platforming with rapid tap shooting and tap-and-hold jumping

The game follows the classic formula that fans have come to expect: an introduction level followed by a stage choose screen featuring eight all-new Maverick bosses. Each boss has its own level, which has classic platforming action, opponent placement, and environmental hazards to keep you on your toes. Defeating a Maverick unlocks a new Special Weapon that can be swapped out as you progress, ideal for whacking the later monsters. Armor upgrades are buried away in each level, and you’ll need to investigate carefully to uncover them all. Collectibles and secrets await anybody who takes the time to venture off the usual path. Oh, there’s also an accomplishment system to keep you going back for more.

The pixel art on display is expertly created to transport you back to the Super Nintendo era. From the characters to the backdrops and even the effects, everything is vintage 16-bit design with a modern gloss that adds just a little more depth without sacrificing any of the original beauty. The soundtrack is impressive, with over 25 brand new tracks to keep the energy up and memories pouring back. The team that worked on this knows Mega Man X inside and out, and it shows.

The controls are as responsive as you’d expect, with X able to dash, wall-jump, charge up his X-Buster, and cycle through his firearms with the same ease you recall. Of course, complete controller functionality is built in from the start, but some users have reported that keyboard input feels smoother right now. Ross has released a fast patch to fix Mac and Linux support, improve Android rotation, and begin addressing all of the comments received from the first wave of players.

Everything runs on the Godot engine, and don’t worry, you get the whole kit from the start. Players will not have to deal with demos or unfinished business. You can already get it on Windows, Android, and Steam Deck, and the Mac version went online shortly after the first patch. The best part is that it’s completely free on itch.io, as Ross says the goal is purely non-commercial and unaffiliated with Capcom. It’s simply something he wanted to make and share with the rest of us to play.

A threat actor is selling employee databases allegedly stolen from the Microsoft Azure infrastructure of multiple Fortune 500 companies after gaining access using compromised credentials.

​Starting July 31st, multiple posts from someone using the alias “TheHatman” advertised data dumps from major organizations, including McDonald’s, Gap Inc., Vodafone, Tata Consultancy Services, HCL Technologies, InterContinental Hotels (IHG), and Kyndryl.

In total, the threat actor claims to have 3.64 million data records, with the most recent breach posted on Sunday, containing an alleged 1.7 million employee records from McDonalds.

“I’m selling McDonald’s Corporation internal employee dump downloaded directly from Azure Tenant using compromised credentials,” the threat actor says in the post.

TheHatman says that the information includes names, employee IDs, email addresses, job titles, phone numbers, postal addresses, service accounts, and other tenant account records.

The second-largest data dump advertised is allegedly stolen from Tata Consultancy: an Azure dump with more than 800,000 employee records “downloaded directly from Azure Tenant using compromised credentials,” the cybercriminal states.

However, in a notification to the National Stock Exchange of India, Tata says it investigated the alleged breach and found no “credible evidence of a breach of TCS systems or customer environments.”

The company states that the details appear to be at least four years old and include only basic employee information.

“The attacker claims to have used password spray and Multi-Factor Authentication (MFA) fatigue as the attack vector. The Company has had strong safeguards in place against such techniques for more than two years,” Tata says.

The company also added that it reviewed its defenses and found that they remain effective.

In a statement for BleepingComputer, a Gap Inc. spokesperson said that the company found no evidence of a breach. Additionally, the advertised data is not sensitive in nature and “dated back to several years ago.”

“Our preliminary investigation indicates that the data in question is limited in scope, non-sensitive and dated back to several years ago. Notably, there is no evidence to suggest that our corporate systems have been compromised,” the Gap Inc. representative said.

Between July 31st and August 16, TheHatman has offered to sell data dumps for the following organizations:

For each advertised database, TheHatman also provided a sample database for potential buyers to verify the data.

Cybercrime intelligence company Hudson Rock analyzed the leaks and confirmed that they contain “foundational corporate directory attributes” and a clear data structure with fields that include “active domains and tenant-specific .onmicrosoft.com structures.”

According to the cybersecurity firm, the dumps also contain service accounts and the names of global administrators, which could facilitate social engineering and spearphishing attacks.

While Hudson Rock has high confidence that the data is authentic, the access vector and exfiltration method remain unknown. BleepingComputer has not been able to independently verify that the data is authentic.

BleepingComputer contacted the listed companies about the potential breach but had not received comments by the time of publication.

Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply.

The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments.

Advertisement

Get the report

Anthropic’s revenue continues to not only grow at an historic pace but also to accelerate. The model maker’s annualized revenue run rate — a projection of a full year’s revenue based on a recent, shorter period —surpassed $65 billion at the end of July, Bloomberg reported on Monday, up from $47 billion in May and just $9 billion at the end of last year.

Anthropic didn’t immediately respond to our request for comment.

The company’s investors expect it to continue to grow at approximately the same rate for the remainder of the year, finishing 2026 between $100 billion and $120 billion, the Financial Times reported.

Meanwhile, rival OpenAI has doubled its revenue to $40 billion, up from $20 billion at the end of 2025, Bloomberg reported last week.

The two companies may calculate their revenue metrics differently, but Anthropic’s growth rate has captivated investors far more than OpenAI’s has.

Both companies have filed confidential IPO paperwork. Anthropic is expected to hit the public markets ahead of OpenAI — possibly as soon as this fall. Anthropic will be seeking a public valuation of $2 trillion or more, according to the Financial Times, which would make it the largest market debut on record.

Anthropic was last valued at $965 billion in late May, when it raised a $65 billion round.

devops

Screen readers suffer as nearly 90% of top websites tested contain HTML spec violations

Are web coders losing their respect for standards? A recent review of the world’s largest sites suggests so.   

ValidateHTML, a project by independent French developer Théo Ducreux, examined the contents of the 5,000 most widely used web domains, as estimated by the Tranco research project. They include the usual suspects (Google, YouTube, Akamai) plus those without public-facing components (Google’s “gstatic,” EZVIZ’s “Ezviz7”).

Ducreux found more than half – 2,656 sites – served a home page readable by humans. Many of the others, such as Google and EZVIZ’s, were just traffic redirection sites, a beacon for a company’s products to call home.

Headless or not, nearly 90 percent of these sites were not serving proper HTML, as specified by the World Wide Web Consortium (W3C) and the WHATWG (Web Hypertext Application Technology Working Group). These standards are important so that different browsers and reading tools can render the requested pages in more or less the same way.  

Ducreux tallied a total of 100,305 HTML violations (including the most recent HTML5 specs) across all these sites, plus an additional 18,863 CSS errors. In other words, 87.2 percent of sites violate web specs in at least one place. Only 12.8 percent have fully valid HTML. And only 2.6 percent of all sites are completely clean, returning zero errors and zero best-practice warnings.

More alarming, over a third of sites failed accessibility checks, which to their owners is more of a legal liability (and hence a compliance issue) than an embarrassment, given the potential reach of the 2025 European Accessibility Act. (Here in the U.S., woe be the accessibility-inconsiderate e-commerce site that raises the ire of the National Federation of the Blind.) 

Accessibility-wise, 20.4 percent of sites are missing alt text on images, which often contain vital information. Worse, 41.6 percent of pages are missing ARIA labels that identify page regions, potentially leaving screen readers struggling to convey the page’s structure. Here’s an iframe, figure it out. 

“Screen readers don’t have anywhere near the error tolerance Chrome has, so markup that looks totally fine to you can be broken for someone using assistive tech,” Ducreux told The Register. 

For the project, he used a home-built web crawler and a set of open-source parsers (HTML-validate, CSS Validator, Lightning CSS, fast-xml-parser). “The site, the scoring, and the crawl logic are my own code,” Ducreux explained by email. Ducreux is the principal maintainer for the project and the site. “No team, no funding, no company.”

An electronic Tower of Babel

Does lack of adherence to standards really matter? It’s an open question.

“The web is more a social creation than a technical one,” web creator Tim Berners-Lee wrote back in 1999. If we all go off and make our own standards, then it’s harder to connect and communicate.

The currently fashionable Dead Internet Theory – in which AI machines and not we meat sticks now make up most internet traffic – won’t get developers off the hook. As Shopify engineers have found, even superintelligence prefers shipshape markup. 

Browsers deserve part of the blame. Unless they have a fetish for the short-lived but brutally unforgiving XHTML, browsers today will just ignore any code they can’t grok. 

“Browsers are so good at error recovery that nothing forces anyone to fix” their code, Ducreux wrote on the site, “which is exactly why it accumulates.” Indulgence encourages slothfulness.  

Front-end frameworks create their own mischief. The single most common failure identified (in more than 59 percent of all the sites) is the misplaced element, aka incorrect tag nesting. Did you know you’re not allowed to put a style element within a div box? Many frameworks do, evidently. 

“It is a nesting problem created at build time, not in an editor,” Ducreux wrote on the site. “Nobody writes that by hand.” 

Many folks believe sloppy HTML is no big deal, including some in the peanut gallery at Hacker News. One contributor noted that “HTML5 defines a method for turning more-or-less any sequence of bytes into the same DOM tree,” DOM being the Document Object Model the browser uses to understand a web page. 

Ducreux himself doesn’t see bad HTML as an existential threat to the web, outside of hindering accessibility, which can leave someone guessing at online content they may really need to understand. 

“Honestly, most invalid HTML doesn’t break anything visible, browsers are built to guess what you meant and just patch it silently,” he wrote by e-mail. But at the same time, the web isn’t just about browsers.  

“AI agents reading pages, voice assistants, translation tools, screen readers – none of them have Chrome’s 20 years of guessing your intent,” he wrote. “Writing to spec is what makes a page mean the same thing to all of them, not just to whoever’s rendering it in Chrome that day.” ®