請同意我們的隱私權規範，才能啟用聽新聞的功能。

（中央社記者吳家豪台北16日電）蘋果公司（Apple）今天公布年度「環境進度報告」，說明邁向2030年整體營運碳中和的進展，並宣布2025年所有出貨產品中，來自再生材料的內容物達到創紀錄的30%。

蘋果營運長沙比汗（Sabih Khan）表示，在業務的各個層面，蘋果都展現創新與合作如何將宏大的想法與大膽的目標轉化為可衡量的成果。從擴大再生材料的使用，到包裝去除塑膠，蘋果正建立新的標準，激勵蘋果持續精進，並為人類與地球的福祉更加努力。

除了去年所有出貨產品達到30%再生材料以外，蘋果去年也在特定零組件的再生材料使用方面達成多項目標。現在所有蘋果設計的電池皆採用100%再生鈷金屬，所有磁石皆採用100%再生稀土元素；所有蘋果設計的印刷電路板，皆採用100%再生鍍金與錫焊料。

蘋果於去年完成轉用100%纖維製包裝，實現2025年包裝去除塑膠的承諾。在過去10年間，蘋果的工程師和設計師開發常見包裝元件的替代方案，將塑膠保護片與托盤以負責任來源紙材製成的版本取代。

在產品中擴大使用再生材料的同時，蘋果也正在開發新技術，將裝置上可回收的材料最大化。蘋果在加州先進回收中心（Advanced Recovery Center），推出一條全新世界級電子產品回收產線Cora，透過精密粉碎與先進感測技術，讓材料回收率顯著高於業界基準。

此外，蘋果拆解機器人Daisy位於荷蘭及美國，目前可拆解36種iPhone型號，回收更多內部材料，每台Daisy每年可處理約120萬台iPhone。另一台機器人Dave專門拆解觸覺引擎Taptic Engine，回收珍貴的稀土磁鐵、鎢及鋼材。

蘋果先前推出的平價筆記型電腦MacBook Neo，是歷來碳排量最低的MacBook，再生材料含量達60%，為所有蘋果裝置最高，包括電池採用100%再生鈷金屬，以及所有磁石採用100%再生稀土元素。（編輯：楊凱翔）1150416

In Brief

More than three years after the emergence of generative AI, AI-assisted coding remains by far the most popular and lucrative use case for the technology.

Although multiple companies — including Anthropic, maker of Claude Code, as well as Cursor and Cognition — are already vying for dominance, investors believe there is room for at least one more player.

On Wednesday, Factory, a startup developing AI agents for enterprise engineering teams, announced it had raised $150 million at a $1.5 billion valuation. The round was led by Khosla Ventures, with participation from Sequoia Capital, Insight Partners, and Blackstone. Keith Rabois, a managing director at Khosla Ventures, joined the startup’s board.

Factory founder Matan Grinberg told the Wall Street Journal that the company’s key differentiator is its ability to switch between different foundation models, such as Anthropic’s Claude or Chinese AI startup DeepSeek. However, startups like Cursor also don’t rely on a single model to generate code.

Factory’s customers include engineering teams at Morgan Stanley, Ernst & Young, and Palo Alto Networks.

The startup was founded in 2023 after Grinberg, then a PhD student at UC Berkeley, cold-emailed Sequoia partner Shaun Maguire. The two bonded over mutual academic interest. (Maguire’s PhD from Caltech is in the same area of physics Grinberg was studying.)

Maguire convinced Grinberg to drop out and launch Factory, with Sequoia backing the startup at the seed stage.

AI video generation startup Luma has launched Innovative Dreams, a production company built in partnership with Wonder Project, a streaming service that produces religious films and TV on Amazon Prime.

The tie-up’s first show will be called “The Old Stories: Moses,” starring British actor Ben Kingsley and set to launch this spring on Prime Video.

“Innovative Dreams is a production services company where seasoned filmmakers from director Jon Erwin’s team and Luma’s creative technologists work with great studios and filmmakers to help them realize ambitious ideas,” Luma said Thursday in a social media post.

The company envisages creative teams collaborating in real time with Luma Agents to make changes to sets, props, and lighting, as well as bring in footage of human actors. Luma Agents are the company’s recently launched tools designed to handle end-to-end creative work across text, image, video, and audio.

“This is a significant improvement over the current virtual production and performance capture processes where things come together only in post,” Luma’s post said. “This is the leverage of AI — not just faster or cheaper, but better than what came before.”

Luma isn’t the only startup to move from tooling to production. AI startup Higgsfield last week launched an original series, starting with a 10-minute sci-fi episode, and London-based creative studio Wonder Studios is working on a documentary with Campfire Studios.

The launch comes the same week that competitor Runway’s co-founder and co-CEO Cristóbal Valenzuela said film studios should take the $100 million they spend on a single film and instead use AI to produce 50 films in order to increase their chances of making a blockbuster.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

Luma founder and CEO Amit Jain has made a similar case, telling TechCrunch that Hollywood’s soaring production costs have made filmmaking increasingly constrained. Generative AI, he argues, could make filmmaking faster, cheaper, and more efficient without sacrificing quality.

That thinking underpins Luma’s new partnership with Wonder Project.

Wonder Project, launched in 2023, is run by director Jon Erwin and former Netflix executive Kelly Hoogstraten with the goal of serving the faith and values audience globally. Their first project, “House of David,” a Biblical drama series about the life of King David, was released on Amazon Prime in 2025.

It’s unclear whether Innovative Dreams will focus solely on religious and faith-based content or expand beyond Wonder’s remit. TechCrunch has reached out for clarification.

In a video promoting the partnership, Erwin said Innovative Dreams will use a new “real-time hybrid filmmaking” process that combines performance capture (as in “Avatar”) and virtual production (as in “The Mandalorian”), done live and more cheaply using Luma’s tools.

Performance capture is a technique where actors perform in a green-screen environment wearing suits and facial markers so their movements and expressions can be digitally captured and turned into animated characters. Virtual production involves actors performing on set, often in front of massive LED screens instead of a green screen while real-time game-engine graphics create the environment around them, blending the physical and digital worlds during the shoot.

Luma’s tools, Erwin said, allow them to film a human actor anywhere and then transport that to a photorealistic scene, or go even further by generating a new face so it looks like a completely different person but still maps onto the actor’s movements and facial expressions.

Reed Hastings, the co-founder and chairman of Netflix, is leaving the company’s board when his term expires this summer. Hastings is stepping aside to focus on “philanthropy and other pursuits,” the company said in a letter to shareholders.

The departure was revealed in the company’s first-quarter earnings report published Thursday. Hastings will officially leave the board when his term is up in June, according to the filing.

“Netflix changed my life in so many ways, and my all‑time favorite memory was January 2016, when we enabled nearly the entire planet to enjoy our service,” Hastings said in a statement included in the earnings report.

“My real contribution at Netflix wasn’t a single decision,” Hastings continued. “It was a focus on member joy, building a culture that others could inherit and improve, and building a company that could be both beloved by members and wildly successful for generations to come.”

The outgoing founder also thanked Ted Sarandos, the company’s CEO, and Greg Peters, its co-CEO.

In 1999, when mom-and-pop video rental stores still coated the country, Hastings and Netflix co-founder Marc Randolph launched a subscription DVD-by-mail business for movies. In the coming years, Netflix would help wipe out many of those mom-and-pop stores (as well as mega-rental chain Blockbuster), as its offerings expanded from a disc-delivery service to digital delivery (the company officially retired its disc-delivery service in 2023). Under Hastings’ leadership, Netflix became a pioneer in the streaming industry, inspiring a wave of similar services from the likes of Amazon, HBO, Disney, Hulu, and others.

“Reed built a culture of innovation, integrity and high performance that defines who we are today,” the company said in the earnings report. “His vision and leadership pioneered how the world is entertained, and his legacy and impact are not only felt by all of us at Netflix, but by audiences around the world.”

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

While Hastings heads for the exit, Netflix said Thursday that it is looking to expand into new technological frontiers, including generative AI. The report also mentions the streaming giant’s recent acquisition of InterPositive, Ben Affleck’s AI company.

The company reported $12.25 billion in revenue in the first quarter, a 16.2% increase from the same period last year. Netflix’s net income rose nearly 83% to $5.28 billion.

In Brief

AI infrastructure company Upscale AI is reportedly in talks to nab its third funding round since launching just seven months ago, according to Bloomberg.

This latest round — which aims to raise around $180 million to $200 million — would value the company at about $2 billion. The company announced a $200 million Series A in January and a $100 million seed round in September, when it first launched. Investors in the company include Tiger Global Management, Xora Innovation, and Premji Invest.

Notably, Upscale AI has yet to release a product. However, it’s said to focus on building custom chips and on the infrastructure to enable them to communicate effectively. The company is betting on a full-stack solution and open standards as being the future of scalable AI infrastructure.

The rumored valuation and raise are part of the startup playbook in this age of AI, where companies grow fast and valuations grow faster, but the hope for the next big thing outpaces it all.

Physical Intelligence, the two-year-old, San Francisco-based robotics startup that has quietly become one of the most closely watched AI companies in the Bay Area, published new research Thursday showing that its latest model can direct robots to perform tasks they were never explicitly trained on — a capability the company’s own researchers say caught them off guard.

The new model, called π0.7, represents what the company describes as an early but meaningful step toward the long-sought goal of a general-purpose robot brain: One that can be pointed at an unfamiliar task, coached through it in plain language, and actually pull it off. If the findings hold up to scrutiny, they suggest that robotic AI may be approaching an inflection point similar to what the field saw with large language models — where capabilities begin compounding in ways that outpace what the underlying data would seem to predict.

But first: The core claim in the paper is compositional generalization — the ability to combine skills learned in different contexts to solve problems the model has never encountered. Until now, the standard approach to robot training has been essentially rote memorization — collect data on a specific task, train a specialist model on that data, then repeat for every new task. π0.7, Physical Intelligence says, breaks that pattern.

“Once it crosses that threshold where it goes from only doing exactly the stuff that you collect the data for to actually remixing things in new ways,” says Sergey Levine, a co-founder of Physical Intelligence and a UC Berkeley professor focused on AI for robotics, “the capabilities are going up more than linearly with the amount of data. That much more favorable scaling property is something we’ve seen in other domains, like language and vision.”

The paper’s most striking demonstration involves an air fryer the model had essentially never seen in training. When the research team investigated, they found only two relevant episodes in the entire training dataset: One where a different robot merely pushed the air fryer closed, and one from an open source dataset where yet another robot placed a plastic bottle inside one on someone’s instructions. The model had somehow synthesized those fragments, plus broader web-based pretraining data, into a functional understanding of how the appliance works.

“It’s very hard to track down where the knowledge is coming from, or where it will succeed or fail,” says Ashwin Balakrishna, a research scientist at Physical Intelligence and a Stanford computer science PhD student. Still, with zero coaching, the model made a passable attempt at using the appliance to cook a sweet potato. With step-by-step verbal instructions — essentially, a human walking the robot through the task the way you might explain something to a new employee — it performed successfully.

That coaching capability matters because it suggests robots could be deployed in new environments and improved in real time without additional data collection or model retraining.

So what does it all mean? The researchers aren’t shy about the model’s limitations and are careful not to get ahead of themselves. In at least one case, they point the finger squarely at their own team.

“Sometimes the failure mode is not on the robot or on the model,” Balakrishna says. “It’s on us. Not being good at prompt engineering.” He describes an early air fryer experiment that produced a 5% success rate. After spending about half an hour refining how the task was explained to the model, it jumped to 95%, he says.

Image Credits:Physical Intelligence

The model also isn’t yet capable of executing complex multi-step tasks autonomously from a single high-level command. “You can’t tell it, ‘Hey, go make me some toast’,” Levine says. “But if you walk it through — ‘for the toaster, open this part, push that button, do this’ — then it actually tends to work pretty well.”

The team also acknowledged that standardized benchmarks for robotics don’t really exist, which makes external validation of their claims difficult. Instead, the company measured π0.7 against its own previous specialist models — purpose-built systems trained on individual tasks — and found that the generalist model matched their performance across a range of complex work, including making coffee, folding laundry, and assembling boxes.

What may be most notable about the research — if you take the researchers at their word — is not any single demo but the degree to which the results surprised them, people whose job it is to know exactly what is in the training data and therefore what the model should and shouldn’t be able to do.

“My experience has always been that when I deeply know what’s in the data, I can kind of just guess what the model will be able to do,” Balakrishna says. “I’m rarely surprised. But the last few months have been the first time where I’m genuinely surprised. I just bought a gear set randomly and asked the robot, ‘Hey, can you rotate this gear?’ And it just worked.”

Levine recalled the moment researchers first encountered GPT-2 generating a story about unicorns in the Andes. “Where the heck did it learn about unicorns in Peru?” he says. “That’s such a weird combination. And I think that seeing that in robotics is really special.”

Naturally, critics will point to an uncomfortable asymmetry here: Language models had the entire internet to learn from. Robots don’t, and no amount of clever prompting fully closes that gap. But when asked where he expects the skepticism, Levine points somewhere else entirely.

“The criticism that can always be leveled at any robotic generalization demo is that the tasks are kind of boring,” he says. “The robot is not doing a backflip.” He pushes back on that framing, arguing that the distinction between an impressive robot demo and a robotic system that actually generalizes is precisely the point. Generalization, he suggests, will always look less dramatic than a carefully choreographed stunt — but it is considerably more useful.

The paper itself uses careful hedging language throughout, describing π0.7 as showing “early signs” of generalization and “initial demonstrations” of new capabilities. These are research results, not a deployed product, and Physical Intelligence has been restrained from the start about commercial timelines.

When asked directly when a system based on these findings might be ready for real-world deployment, Levine declines to speculate. “I think there’s good reason to be optimistic, and certainly it’s progressing faster than I expected a couple of years ago,” he says. “But it’s very hard for me to answer that question.”

Physical Intelligence has raised over $1 billion to date and was most recently valued at $5.6 billion. A significant part of the investor enthusiasm around the company traces to Lachy Groom, a co-founder who spent years as one of Silicon Valley’s most well-regarded angel investors — backing Figma, Notion, and Ramp, among others — before deciding that Physical Intelligence was the company he’d been looking for. That pedigree has helped the startup attract serious institutional money even as it has refused to offer investors a commercialization timeline.

The company is now said to be in discussions for a new round that would nearly double that valuation figure to $11 billion. The team declined to comment.

For a successful deep tech startup, the laboratory breakthrough is only the first step. To spin out an innovative technology into a scalable business, a founder has to navigate a complicated legal process.

This week on Build Mode, Isabelle Johannessen speaks with Capella Kerst, founder and CEO of geCKo Materials and the 2024 TechCrunch Startup Battlefield runner up. Kerst was a Stanford PhD student, working on bio-inspired adhesives — materials modeled after the microscopic hairs that allow geckos to stick to walls. She wasn’t looking to be a founder but when she had a major breakthrough that made it possible to make the material rapidly and reliably, she knew it was a viable product.

But translating a lab discovery into a startup is less about the “eureka” moment and more about everything that comes after.

“I got up the courage to really address my advisor in a very like we need to have a serious conversation about me spinning out this company and starting it,” Capella said. And that conversation was the beginning of a long process of building geCKo materials.

Here is a roadmap for founders spinning a product out of academia.

Reach out to prior contributors. Connect with all the people who’s work lead up to the big breakthrough. Kerst offered these contributors the options to join the company, become advisors, or receive compensation.

Connect with all the people who’s work lead up to the big breakthrough. Kerst offered these contributors the options to join the company, become advisors, or receive compensation. Ensure you reach the requirements for the licensing process. Kerst started conversations with Stanford’s office of Technology licensing early and she learned to spin it out, she’d need to complete her PhD. These early conversations also helped her prepare for the process to come.

Kerst started conversations with Stanford’s office of Technology licensing early and she learned to spin it out, she’d need to complete her PhD. These early conversations also helped her prepare for the process to come. Form the company and lawyer up. In this case, Stanford provided a list of lawyers Kerst had to use. “I spent time interviewing lawyers and finding out about case studies, what other people got, what did you push, what didn’t work, what got pushed back on. And so I just learned a ton, picked my lawyer, and then we went at Stanford.” said Kerst, “I was like, I want a good deal, because I want this to be a big company, and I think it’ll benefit everybody.”

In this case, Stanford provided a list of lawyers Kerst had to use. “I spent time interviewing lawyers and finding out about case studies, what other people got, what did you push, what didn’t work, what got pushed back on. And so I just learned a ton, picked my lawyer, and then we went at Stanford.” said Kerst, “I was like, I want a good deal, because I want this to be a big company, and I think it’ll benefit everybody.” Transition fully into the founder role. Once the licensing agreement is finalized, it’s time to shift from PhD to CEO.

Five years later, geCKo Materials is continuing to scale and develop new ways to apply its adhesive technology which is being tested in applications ranging from robotics and manufacturing to automotive and even space. The company’s material is already in use on the International Space Station, and Kerst says the long-term vision includes replacing traditional attachment methods like Velcro or suction systems.

Loading the player…

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

Apply to Startup Battlefield: We are looking for early-stage companies that have an MVP. So nominate a founder (or yourself): techcrunch.com/apply. Be sure to say you heard about Startup Battlefield from the Build Mode podcast.

TechCrunch Disrupt: If you’re thinking about applying to Startup Battlefield, then October 13 to 15 in San Francisco, we’re back for TechCrunch Disrupt, where the Startup Battlefield 200 takes the stage. So if you want to cheer them on, or just network with 1000s of founders, VCs, and tech enthusiasts, then grab your tickets.

Use code buildmode15 for 15% off any ticket type.

Isabelle Johannessen is our host. Build Mode is produced and edited by Maggie Nye. Audience Development is led by Morgan Little. And a special thanks to the Foundry and Cheddar video teams.

In Brief

Slash Financial, which offers business banking accounts, corporate credit cards, transfers, and crypto, has raised a $100 million Series C round at a $1.4 billion valuation from a crowd of A-list investors, the company announced.

Fintech-focused Ribbit Capital, Khosla, and Goodwater Capital led the round. Returning investors NEA and Y Combinator also invested.

Slash was founded about five years ago by CEO Victor Cardenas and CTO Kevin Bai, both who were 19 years old at the time, and are now 24, Bloomberg reports. They dropped out of college to build a fintech company focused on sneaker resellers. When the startup’s main customer, Yeezy, hit trouble after its founder Kanye West spewed antisemitic remarks, they pivoted to focus on a few verticals.

Now, the startup has become generalist, not targeting any one industry, Cardenas recounted in his blog post about the raise. He said the company is generating $300 million in annualized revenue, profitably, and claims 5,000 companies as customers. Even with such growth numbers, Slash has plenty of competition, including $32 billion valuation Ramp and Brex, which was recently acquired by Capital One.

There is currently a low-grade war between OpenAI and Anthropic over who can release the most convenient and powerful AI-coding tools and, so far, Anthropic seems to be winning. Claude Code has been dubbed the tool of choice for many businesses, as TechCrunch reported last week, but OpenAI isn’t giving up yet.

This week, OpenAI announced a revamp of Codex, its own automated tool, with a variety of new updates designed to give it significantly expanded powers.

On Thursday, the company announced a plethora of new features and updates, perhaps the most notable of which is that Codex can now operate in the background on your computer — opening any app on your desktop and carrying out operations with a cursor that clicks and types.

Functionally, what this does is allow Codex to deploy multiple agents, all of which work on a user’s Mac “in parallel, without interfering with your own work in other apps,” the company said in a blog post. In other words, because of the way Codex runs in the background, a user can still be using the machine as the agent goes about its own work. The agent will then function, according to the company, as a kind of coding buddy that does auxiliary tasks while you work on topline projects. OpenAI’s lists “iterating on frontend changes, testing apps, or working in apps that don’t expose an API” as potential use cases for this kind of agentic assistance.

Overall, this agentic update and other new additions demonstrate OpenAI’s desire to not only make Codex a competitive coding assistant but also a more multifaceted tool that can be integrated into a variety of corporate workflows.

Watchers of the AI coding space will also note that some of the powers OpenAI is now adding to Codex seem to resemble those previously released by Anthropic for Claude Code. Last month, Anthropic announced that Claude and Cowork could remotely control your Mac and desktop on a user’s behalf while they were away from their keyboard.

In addition to the agentic tools, OpenAI’s Codex now has an in-app browser, which allows a user to issue commands to the agentic tool, which it will then ostensibly carry out on specific web applications. OpenAI says this function will be useful for front-end and game development, and that it plans to eventually expand the capability so that Codex can “fully command the browser beyond web applications on localhost.”

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

There are other updates. A new feature in preview called “memory” allows Codex to recall previous work sessions and generate important context about how a particular user works. The agent has also been given a new image-generation ability, which OpenAI says can be used to create product concepts, slide visuals, mockups, placeholder images, and other corporate paraphernalia. Finally, to expand Codex’s ability to get things done, the company has announced 111 plug-in integrations from apps like CodeRabbit and GitLab Issues, which allows Codex to carry out tasks involving those tools.

The way OpenAI has framed it, these plug-ins give Codex the ability to carry out minor clerical work to organize your work life. For example, if you want Codex to take a look at your Slack channels and Google calendar and give you a to-do list for a given day, OpenAI says that it can now do that for you.

A new pay-as-you-go Codex pricing option for ChatGPT enterprise and business customers has also been announced in an apparent effort to give users more flexibility when it comes to procuring the coding tool’s services.

Once considered the undisputed leader of its industry, OpenAI has more fiercely competed with Anthropic in recent months, with a focus on enterprise capabilities and a retreat from consumer tools like its social video app Sora 2. The company has also battled various controversies in recent months, including lawsuits over ChatGPT’s alleged mental health impact on some users.

In Brief

A coalition of global law enforcement agencies have sent emails to more than 75,000 alleged cybercriminals who paid for a service to launch cyberattacks that can knock websites offline.

On Thursday, Europol announced the coordinated operation against several distributed denial-of-service (DDoS) for-hire services, which allow criminals to launch cyberattacks without needing to have any hacking skills, nor the need to run their own infrastructure.

Part of the law enforcement action — dubbed Operation PowerOFF — included Europol sending warning emails and letters to more than 75,000 people who are suspected of using these DDoS-for-hire services.

Europol said it obtained information about the alleged cybercriminals by raiding and seizing servers associated with these services, allowing the police to identify their registered users.

The action also resulted in four arrests, the takedown of 53 domains, and police executing 24 search warrants.

DDoS attacks remain relatively common for their ability to cause disruption while being relatively easy to carry out, thanks in part to for-hire services. Last year, Cloudflare said it mitigated what it called the largest DDoS attack to date, which registered a peak of 29.7 terabits per second. In the last few years, the FBI has conducted several operations against DDoS-for-hire services.

In Brief

Mike Krieger, Anthropic’s chief product officer, resigned from the board of interface design company Figma on April 14.

His departure was disclosed to the U.S. Securities and Exchange Commission by the publicly traded $10 billion company the same day that The Information reported Anthropic’s next model, Opus 4.7, will include design tools that could compete with Figma’s primary offering.

Figma is the developer of a popular tool for user experience designers who build interfaces for websites and apps. The company has collaborated closely with Anthropic to integrate the frontier lab’s AI models into its products as assistants for its users.

Krieger, who previously co-founded Instagram and the AI-powered news app Artifact, became the top product executive at Anthropic in 2024 and joined the board of Figma less than a year ago.

Krieger’s departure and any forthcoming design tools will be another data point for investors who fear the SaaSpocalypse — that the largest AI labs will come to dominate software businesses, a thesis that has rocked public markets at times this year. For example, iShares’s primary software ETF, IGV, is down nearly 18% this year.

Anthropic, meanwhile, is turning down investors who want to buy into the company at $800 billion — more than double the valuation from its most recent round at the beginning of the year.

But companies like Anthropic and OpenAI still have to prove their ultra-capable models can truly replicate the domain experience and relationships of established software brands. Figma’s stock price is up 5% since Krieger’s departure was disclosed, though we’ll see what happens with the next Opus release.

Roblox is introducing new agentic features to help developers plan, build, and test games on its platform, the company told TechCrunch exclusively. Roblox is revamping Roblox Assistant, its plain-language AI tool for game development, to help creators throughout the entire development process.

The company says that AI tools that take in a prompt and output a solution in one step can often fail to truly capture a creator’s original intent. That’s why it’s introducing an enhanced “Planning Mode” that transforms Assistant into a collaborative partner that can analyze a game’s code and data model, ask clarifying questions, and translate prompts into editable action plans.

Planning Mode helps developers create a plan for their game, get feedback to refine details, finalize the approach, and then implement that plan. Creators can tweak the plan and add context to ensure their intent is clearly reflected before any changes are made.

Image Credits:Roblox

For example, if a creator tells Assistant to “create a park mini game with a fountain and foliage where characters have to collect coins,” the Assistant may ask what visual style they want the park to have, with options like cartoony, realistic, and fantasy. Or, the Assistant may ask how they want the park’s assets, like a fountain and foliage, to be created, offering options such as building from scratch, using models from the Creator Store, or a mix of both.

Once there is a plan in place, Planning Mode will leverage Roblox’s other AI tools while it creates the game. These AI tools include two new ones announced today: Mesh Generation and Procedural Model Generation, which are designed to speed up development.

Mesh Generation makes it easy to add fully textured meshes, or 3D objects, directly into the game world. Roblox says that during the early stages of development, developers often create placeholder assets to understand how the player will interact with the world. With Mesh Generation, creators can quickly create 3D models instead of having to rely on low-quality placeholders.

Image Credits:Roblox

For example, creators can ask Assistant to generate a campfire, then add light to make it more realistic, and then set the scene at night.

Roblox will also soon introduce “Procedural Models” to allow developers to create editable 3D models with code and Assistant. Because Assistant understands 3D space and physical relationships, creators can use prompts to place and scale objects based on other objects in the scene.

Attributes like the number of shelves in a bookcase or the height of a staircase can be adjusted dynamically, creating editable building blocks that can be refined and reused elsewhere.

“The launch of our agentic features in Roblox Studio reduces barriers between creative vision and execution, said Nick Tornow, senior vice president of Engineering, in a statement to TechCrunch. “Creating with Planning Mode and our Procedural Generation tools is a powerful new method for creators to turn their concepts into gameplay. Assistant works as a multi-step, collaborative development partner — accelerating the process of planning, building, and testing, so creators can get from idea to reality faster.”

As Planning Mode executes against the plan, it will use playtesting tools to read output logs, capture screenshots, use inputs like a keyboard and mouse to check design and gameplay, and identify bugs and provide feedback to the Assistant so it can fix them automatically.

Image Credits:Roblox

“With the new capabilities across planning, building, and testing, Assistant is better at using agentic loops to test different aspects of the game, surface suggested solutions, and then incorporate the results into future planning loops, creating a self-correcting system that becomes more accurate over time,” Roblox explained in a blog post.

Roblox also announced that it’s working on enabling multiple AI agents to work together in parallel, run long, complex workflows in the cloud, and handle tasks like coding, testing, and creating more realistic game characters. It also wants to ensure creators can seamlessly use Claude, Cursor, Codex, and other third-party tools with Roblox Studio.

Last year, I was telegraphed a subliminal mandate from the indie rock powers that be: I was supposed to like Geese. The young Brooklynites make good music, but are they the saviors of rock and roll, the defining rock band of Gen Z, the second coming of The Strokes?

The buzz around the band would suggest so. After their album “Getting Killed” came out in September, the band was unavoidable if you’re the kind of person who refers to concerts as “shows.” When frontman Cameron Winter played an “extremely sold-out” solo set at Carnegie Hall, people in the audience seemed convinced that they’d be able to look back on that night in 50 years and tell their grandchildren that they witnessed a seminal moment in American musical history — the birth of the next Bob Dylan. How could anyone live up to that hype?

That’s why, when Wired reported that Geese’s popularity was a psyop, I felt vindicated — I was right! I knew it! I was smarter than everyone for only casually enjoying Geese!

But it’s never that simple. The real story is that Geese worked with a marketing firm called Chaotic Good, which creates thousands of social media accounts designed to manufacture trends on behalf of their clients, which also include TikTok favorites Alex Warren and Zara Larsson. This revelation has inspired a range of reactions, from feelings of betrayal to confusion at why anyone is mad about a band doing marketing, a normal thing that bands do.

“On TikTok, it’s really easy to get views. You just post trending audios. But artists can’t do that, because they want to promote their own music,” explained Chaotic Good co-founder Andrew Spelman in an interview with Billboard. “So a big part of what we are doing is posting enough volume across enough accounts with enough impressions to try to simulate the idea that the song is trending or moving.”

When you learn how prevalent these marketing strategies are, it kind of feels like you’re a kid who just learned that the Tooth Fairy isn’t real — you probably had a hunch that something was up, but you want to believe in the fantasy that a fluttering fae is sneaking into your room, and every viral success story is a fairy tale.

It’s not just the music industry taking advantage of this marketing strategy — young startup founders are following the same playbook.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

While preparing for an interview with the Gen Z founders of the fashion app Phia, I searched TikTok to see what real people were saying about the app. I found videos repeating the same talking points about how Bill Gates’ daughter created an app that helps you save money on luxury products, or how using Phia is like having a personal shopping assistant that wants you to get the best deals. When I clicked on these accounts, I found that many of them only ever posted videos about Phia.

It’s not like I caught Phia in some “gotcha” moment. Founders Phoebe Gates and Sophia Kianni aren’t trying to hide their social media strategy — this is just how marketing works now.

“One thing we’ve been trying lately is basically running a creator farm, so we have a ton of different college students that we pay to make videos about Phia on their own accounts,” Kianni said on her podcast. “This is an approach that’s really focused on volume. We have like ten creators, they post twice a day, and we ultimately reach like 600 videos total.”

On TikTok-like feeds, people watch videos in a vacuum, separate from the rest of a creator’s account. Few viewers will stop to look at what else that person is posting, so they won’t suspect that the post about this cool new app could be an inorganic promotion.

Creators will similarly pay armies of teenagers on Discord to make clips of their streams and post them en masse.

“That’s been going on for a bit,” Karat Financial co-founder Eric Wei told TechCrunch last year. “Drake does it. A lot of the biggest creators and streamers in the world have been doing it — Kai Cenat [a top Twitch streamer] has done it — hitting millions of impressions … If it’s algorithmically determined, clipping suddenly makes sense, because it can come from any random account that just has really good clips.”

Marketing firms like Chaotic Good scale that same approach — instead of paying college students or teenage fans to make videos, they buy hundreds of iPhones and make a bunch of social media accounts that they can use to fabricate a viral trend. Spelman told Billboard that Chaotic Good’s office is “overrun with iPhones,” and that they have so many phones that they’re treated like VIPs at Verizon.

“Unfortunately, a lot of the internet is manipulation … Everything on the internet is fake. One thing that we always say is all opinions are formed in the TikTok comments,” Chaotic Good co-founder Jesse Coren noted.

This is the same line of thinking that fuels the Dead Internet Theory, which argues that bot-generated content dominates the web.

If Chaotic Good’s content armies aren’t posting trending audio, they’re commenting on posts about the company’s clients to control the narrative. Instead of waiting to see how fans will respond to a new song, they can use their accounts to flood the comments of videos and talk about how much they love the song.

For Geese, it’s an insult to be called an industry plant. After songwriter Eliza McLamb wrote the blog post that first connected Geese and Chaotic Good, the firm removed mention of Geese and “narrative campaigns” from its website. (The company told Wired that it did this to protect artists from being “wrapped up in false accusations or misconceptions about how their music was discovered.”)

But like the unapologetic marketing behind some Gen Z startups, the global girl group Katseye has been incredibly clear that they are the definition of industry plants — there’s literally a Netflix docuseries, “Pop Star Academy,” that illustrates how a room full of global record executives turned these six young women into superstars, even pitting potential members against each other in a surprise K-pop-style survival show.

When “Pop Star Academy” came out, I watched it in a state of horror — HYBE and Geffen treated these aspiring teenage pop stars like cattle to mold into human billboards that they could use to sell Erewhon smoothies and hair serums. But over the course of the eight-episode series, I became deeply invested in these girls’ lives. I wanted to watch them thrive in the face of unrelenting industry pressure.

I’m sure that this is exactly what Katseye’s management wanted from the documentary — to cultivate a fervent sense of support and defensiveness over the girls, even if it means painting the executives as the bad guys. Fast-forward a few years, and Katseye is performing a song called “Gnarly” at the Grammys — a track that fans hated at first until, suddenly, they didn’t.

It’s hard not to think about Chaotic Good’s “narrative campaigns,” flooding comment sections to control discourse. Though I hated “Gnarly” when it came out, I decided over time that it’s actually an avant-garde masterpiece. Did I change my mind on my own, or was it changed for me? For as much pride as I took in resisting the hype around Geese, I am so wrapped up in Katseye that I’ve spent hours speculating on Reddit forums about the real story behind Manon’s hiatus.

Maybe Geese is a psyop, and maybe Katseye is an industry plant, but do we actually care?

This is not a rhetorical question. The Geese discourse (which could also be manufactured, now that I think about it!) has inspired such varied responses because we have not established clear social norms around what is necessary marketing and what is inauthentic growth hacking.

We, the fans, get to decide now where we draw the line.

Google announced on Thursday that it’s rolling out a new way to explore the web with AI Mode, its conversational search experience. Now, when you’re using AI Mode on Chrome desktop, clicking a link will open the web page side-by-side with AI Mode.

The goal is to make it easier to explore relevant websites, compare details, and ask follow-up questions while preserving the context of your search, the tech giant says.

For example, if you want to purchase a new a coffee maker, you can describe what you’re looking for in AI Mode and get a range of options. Once you click on one, you can open the retailer’s website alongside AI Mode and ask specific questions, like “how easy is this to clean?” AI Mode will then use context from the page and from across the web to answer your questions.

Image Credits:Google /

“Our early testers loved that they didn’t have to constantly switch tabs to get help with a comprehensive article or a long video,” Google explained in a blog post. “And they found that having both Search and the web side-by-side helped them stay focused on their tasks while exploring useful web pages.”

Google also announced a new way to search across the Chrome tabs you’re already looking at.

On Chrome desktop or mobile, you can tap the new “plus” menu in the search box on the “New Tab” page or in AI Mode, then select recent tabs to include them in your search. This means you can mix and match multiple tabs, images, or files and bring that context into your AI Mode searches.

For example, if you’re researching local hiking trails and already have a few tabs open, you can add them to your search and ask for similar trails in a different location. Or, if you’re studying for a statistics exam, you can bring in context from open tabs, class notes, lecture slides, and more to ask for examples to illustrate a concept.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

The new updates to AI Mode are now available in the U.S. Google plans to expand them to additional regions in the future.

Two U.S. citizens were sentenced to seven and a half years and nine years in prison for their roles in a scheme to help the North Korean government place remote IT workers in American companies.

On Wednesday, the U.S. Department of Justice announced the sentencing of Kejia Wang and Zhenxing Wang, both New Jersey residents. The two were accused of providing infrastructure for the fraudulent scheme, in particular for running or managing so-called “laptop farms” inside the U.S., which allowed North Koreans to connect to the laptops and appear like they were living and working in the country.

The scheme netted North Korea around $5 million. It also involved co-conspirators stealing the identities of more than 80 Americans and obtaining work at more than 100 U.S. corporations, including some Fortune 500 companies, according to the DOJ. That also allowed North Korean IT workers not only to get a salary, but also in some cases to steal trade secrets and source code, the Justice Department said.

“The ruse placed North Korean IT workers on the payrolls of unwitting U.S. companies and in U.S. computer systems, thereby harming our national security,” John A. Eisenberg, the DOJ’s assistant attorney general for National Security, was quoted as saying in the announcement.

Prosecutors said that between 2021 and 2024, working with co-conspirators, Kejia Wang oversaw the operation of laptop farms made of hundreds of computers, while Zhenxing Wang hosted laptops at his home. The two also created shell companies with financial accounts linked to the fake IT workers to funnel payments amounting to millions of dollars, which were later transferred overseas. “In exchange for their services, Kejia Wang, Zhenxing Wang, and the four other U.S. facilitators received nearly $700,000 for their respective roles in the scheme,” read the DOJ’s announcement.

In one case, according to the DOJ, the fake IT workers were able to steal data under export control from an unnamed California-based AI company.

The U.S. government also announced rewards of up to $5 million for information that could help counter these schemes, including for data on nine individuals who allegedly worked with Kejia Wang and Zhenxing Wang.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

This is the latest legal action against North Korea’s wide-ranging scheme that has allowed fake IT workers to be hired by hundreds of American and Western companies. Along with major crypto thefts worth more than $2 billion just last year, the North Korean government uses this type of fraud to fund its regime and weapons’ program, which is under heavy sanctions that isolates it from much of the world’s economy.

To counter this threat, some companies and recruiters have come up with inventive strategies, such as asking suspected North Koreans to insult Kim Jong-Un, which is illegal in the country. In a recent viral video of a job interview, the applicant can be seen fumbling after the interviewers asked him to say “Kim Jong Un is a fat ugly pig.” He eventually hung up the call.

The role of observability tools has evolved once again. While the market for solutions to ensure tech systems’ reliability has grown over the years, the center of gravity has steadily shifted from “track everything” to “control complexity and costs.” Meanwhile, the rapid influx and adoption of AI agents within enterprises have only added a brand-new category of workload that needs to be observed.

InsightFinder AI, a startup based on 15 years of academic research, is no stranger to this problem.

The company has been using machine learning to monitor, identify, and proactively fix IT infrastructure issues since 2016 and is now attacking today’s AI model reliability issue with an AI agent solution that can do everything from detection and diagnosis to remediation and prevention.

The company, founded by CEO Helen Gu, a computer science professor at North Carolina State University who previously worked at IBM and Google, recently raised $15 million in a Series B round led by Yu Galaxy, TechCrunch has exclusively learned.

According to Gu, the biggest problem facing the industry today is not just monitoring and diagnosing where AI models go wrong; it’s diagnosing how the entire tech stack operates now that AI is a part of it.

“In order to diagnose these AI model problems, you need to actually monitor and analyze the data, the model, and the infrastructure together,” Gu told TechCrunch. “It’s not always a model problem or a data problem; it’s a combination. Sometimes, it’s simply your infrastructure.”

Gu explained how that looks in real life with an anecdote: One of its customers, a major U.S. credit card company, saw that one of its fraud-detection models was drifting. Because InsightFinder was monitoring all of the company’s infrastructure, it was able to identify that the model drift was caused by outdated cache in some server nodes.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

“The biggest misconception is that AI observability is limited to LLM evaluation during the development and testing phases. On the contrary, a sound AI observability platform should provide end-to-end feedback loop support covering the development, evaluation, and production stages,” she said.

InsightFinder’s newest product, dubbed Autonomous Reliability Insights, can do all this by using a combination of unsupervised machine learning, proprietary large and small language models, predictive AI, and causal inference. This base layer is data agnostic, per Gu, which lets the system ingest and analyze entire data streams to gather signals that can then be correlated and cross-validated to arrive at a root cause.

Now, the observability space is crowded with contenders for a share of the new market that’s been opened up by the influx of AI tools. Nearly a decade into its journey, InsightFinder has been going up against the likes of Grafana Labs, Fiddler, Datadog, Dynatrace, New Relic, and BigPanda, who are all building capabilities to deal with the new problems presented by AI tools.

But Gu isn’t fazed. On the contrary, she claims that InsightFinder’s expertise, experience, and customizability act as a sufficient moat. “We actually rarely lose [customers] to anybody so far … This is about the insights, right? The problem is that a lot of data scientists understand AI, but they don’t understand the system. And a lot of SRE [site reliability engineering] developers understand the system, but not the AI … They don’t look at it, and they don’t understand the intrinsic relationships.”

Today, InsightFinder’s roster of customers includes UBS, NBCUniversal, Lenovo, Dell, Google Cloud, and Comcast. Gu attributes the success to 10 years of working to understand what large enterprise customers need.

“It has come down to working with our Fortune 50 customers to polish and understand the enterprise environment requirements to deploy these kinds of models,” she said. “We have been working with Dell to deploy our AI systems across the world at some of the largest customers we have. This is not something that you can take a foundational AI and just slap on the machine data to do.”

Gu said the company’s revenue stream is “strong,” having grown “over threefold” in the past year. In fact, she says InsightFinder wasn’t looking to raise this Series B at all, and investors approached the company after it won a seven-figure deal with a Fortune 50 company within three months.

InsightFinder will use the fresh capital to make its first sales and marketing hires to expand its team of fewer than 30 people and invest in its go-to-market motion. The company has so far raised a total of $35 million.

As of March, AI traffic to U.S. retailers’ websites rose by 269% over the previous 12 months, continuing the momentum during the holiday shopping season when AI traffic was up by 693%, according to new data released on Thursday by Adobe.

And in the first three months of 2026, AI traffic had risen 393% compared to a year earlier, as more consumers used AI assistants for online shopping.

The change in traffic sources isn’t the only impact. AI visitors are converting better, engaging at higher rates, spending more time on sites, and driving higher revenue per visit, the data shows, often reversing trends from only a year ago, when regular customers were worth more to retailers.

Image Credits:Adobe

Adobe’s insights are based on its analysis of online transactions, via its Adobe Analytics division, which covers over 1 trillion visits to U.S. retail sites. The analysis also relied on a survey of over 5,000 U.S. respondents about their use of AI when shopping, as well as the company’s new AI Content Visibility Checker tool, designed to test retail websites for accessibility by LLMs.

In Adobe’s survey, 39% people said they used AI for online shopping, and 85% said it improved their experience. These findings are likely due to how AI helps people narrow down products to find what they need, and tap into discounts. In addition, 66% of those surveyed said they now believe AI tools provide accurate results when shopping.

Unlike publishers, where AI is causing referral traffic to decline, retailers are incentivized to make their sites AI-friendly.

Adobe’s data found that AI traffic converted 42% better than living, breathing customers in March 2026, setting a new record. Notably, it’s a reversal of a trend that told a different story only a year ago: In March 2025, AI traffic converted 38% worse than regular people.

Image Credits:Adobe

In addition, Adobe found that when a consumer lands on a retail site via an AI source, their engagement rate tends to be 12% higher than those who used non-AI sources. Shoppers also spend more time on the website (48% longer) and browse more pages (13% more pages per visit), the data shows.

In terms of the top line, AI-driven revenue per visit (RPV) was 37% higher than non-AI traffic as of March. Just 12 months ago, regular human traffic was worth 128% more than AI.

However, not all sites are ready for AI, Adobe warned. It found that roughly a quarter of the content on retailers’ homepages has not been optimized for LLMs, nor has the content on category pages. Individual product pages fare even worse: around 34% of pages can’t be properly accessed by AI.

The company suggests that retailers work to make their sites more accessible to LLMs if they want to stay top-of-mind with online shoppers going forward.

Google on Thursday announced that Gemini’s Personal Intelligence feature will get Nano Banana-powered image generation to create images with personalized context.

That means its AI images can be created using Gemini’s understanding of your likes and interests, without those having to be explicitly noted in the prompt.

This works because Gemini already has context of your data through Google account connections, such as Gmail and Google Photos. So, instead of typing “Generate an image of my dream home, my interests are tennis and music,” you can now just say, “Design my dream home.”

Image Credits:Google

What’s more, the Nano Banana-powered connection can also use the labels in your Google Photos, so that it understands names and words that describe a group, like “Family.” For instance, you can create an image by saying, “Generate an image of my family and me doing our favorite activity.”

The company said the “sources” button will show how Gemini derived the context for image generation.

Image Credits:Google

Google said that just like other connections, Gemini might get the context wrong, and you can provide feedback. Plus, you can also provide reference photos for image generation by clicking the “+” icon.

The image-generation feature will be available to Plus, Pro, and Ultra subscribers in the U.S. within the coming days. Google plans to bring the feature to Gemini in Chrome desktop and to other users soon.

Google first launched Personal Intelligence earlier this year and made it available to all U.S. users in March. Earlier this week, the company expanded the feature to more users in countries like India and Japan.

Google said Thursday it blocked a record 8.3 billion ads globally in 2025 — up from 5.1 billion the year before. But the company suspended far fewer advertiser accounts than that surge might suggest, raising questions about how it polices its platform.

The search giant attributed the disparity to its growing use of AI, particularly its Gemini models — Google’s family of AI systems — which Google says allow it to detect and block policy-violating ads earlier and with greater precision. Its AI-driven systems caught more than 99% of such ads last year before they were shown to users, the company said.

Both findings come from Google’s 2025 Ads Safety Report and together reflect a broader change in enforcement. While more problematic ads are being stopped, fewer advertiser accounts are being suspended — suggesting a growing emphasis on blocking individual ads.

Google said the rise in blocked ads also reflects the growing use of generative AI by scammers to produce deceptive content at scale, with its Gemini models helping detect patterns across large campaigns and block them earlier.

The shift also mirrors a wider push by Google to integrate its Gemini models more deeply into its core products and infrastructure, including advertising, where the company is increasingly using AI to automate campaign creation, detect policy violations, and respond to emerging threats in real time.

Image Credits:Jagmeet Singh / TechCrunch

Among the blocked ads and suspended accounts, 602 million ads and 4 million advertiser accounts were linked to scams, the company said.

Google removed over 1.7 billion ads and suspended 3.3 million advertiser accounts in the U.S. in 2025, with ad network abuse, misrepresentation, and sexual content among the most common violations. In India, Google’s largest market by users, it blocked 483.7 million ads — nearly double the previous year — even as account suspensions fell to 1.7 million from 2.9 million, with trademarks, financial services, and copyright issues among the top violations.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

At a virtual briefing, Keerat Sharma, VP and general manager of ads privacy and safety at Google, told reporters the company has shifted toward more targeted, AI-driven enforcement “at a much more granular level, on a creative level, as opposed to using a much more blunt instrument, like advertiser suspensions.” He added that the approach has helped reduce incorrect suspensions by 80% year over year.

Google’s layered defenses, including advertiser verification (a process that requires businesses to confirm their identity before running ads), are designed to prevent bad actors from creating accounts in the first place, Sharma said, adding that this has contributed to the decline in suspensions.

The numbers, Sharma said, are likely to fluctuate over time as Google rolls out new defenses and bad actors adapt, with the company aiming to stop harmful ads as early in the pipeline as possible.

You’ve doubtless heard of hybrid cars, but what about a hybrid cement or glass plant? Likely not, since nearly all of them today run on fossil fuels. But that might change soon — one startup has developed a way to incorporate electric heat into existing facilities. And like a hybrid car, it lets companies save money while also using fewer fossil fuels.

“We hybridize industrial processes,” Carlos Ceballos, co-founder and CEO of NOC Energy, told TechCrunch. “Most companies are willing to electrify, but they don’t want to get rid of fossil fuels yet. In the energy transition, they want to have the opportunity to choose the lowest cost.”

NOC has developed a form of electric heating that can bolt onto existing fossil fuel-fired facilities. Heat from its system can be piped into a glass kiln or various parts of the cement production process. If the cost of electricity rises, the operator can switch it off NOC’s equipment and rely solely on fossil fuels.

Perhaps more important, the startup can deliver heat at temperatures reaching 1,200˚ C, and Ceballos said the company is working toward 1,500˚ C. Those temperatures have been hard to achieve with anything other than fossil fuels or hydrogen, the latter of which is currently too expensive in its non-polluting form. The field doesn’t have many entrants so far — Startup Battlefield alumnus Electrified Thermal Solutions stands out as one potential competitor.

NOC recently raised a $2.7 million seed round, the company exclusively told TechCrunch. The round was led by 360 Capital with participation from SOSV and Desai VC.

NOC’s first customers will likely choose the hybrid format, though they won’t have to if they don’t want to. NOC’s system can store heat for hours on end, allowing companies to use more electricity when the price is cheap — like when the wind is howling or the sun is beating down — and draw on the stored heat when the price spikes.

There are a few elements to NOC’s system that make both hybridization and electricity price arbitrage possible. The first is the induction heating element, which is similar in concept to induction stoves installed in kitchens around the world. Induction heaters use metal coils, usually made of copper, to produce magnetic fields when electricity courses through them. Those magnetic fields cause atoms in certain nearby metals, like steel, to vibrate quickly, generating heat.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

A cutaway view shows the induction coils and steel spheres of the NOC Cell Image Credits:NOC Energy

In NOC’s system, the induction coils go to work on steel spheres that are packed inside large ceramic containers that are 2.5 meters (around 8 feet) across. Wrapped around the containers are the heat-inducing copper coils and heat-conserving insulation. When heat is needed, electricity flows through the copper and warms the steel spheres. Air blown through the spheres extracts the heat and delivers it to where it’s needed, like a glass kiln or a section of a cement plant.

Now, NOC’s approach isn’t the only way to make heat that hot using electricity. Resistive heaters, similar to those found in toasters, can get the job done. But the higher the heat, the shorter their lifespan. At 1,000˚ C, specialized resistance heaters only last around 12 months, Ceballos said. At 1,200˚ C, that drops to three months, he added.

NOC’s heating elements — the copper coils — sidestep that problem entirely because they never come into contact with the heat they generate. That’s the beauty of induction heating — the coils are embedded in half a meter (about 20 inches) of insulation, staying at room temperature while casting their electromagnetic waves inward to the steel spheres.

The insulation is thick enough that NOC’s system can store heat for hours on end. The startup can size the system based on how long a customer wants to store the heat. For longer durations, NOC will stack more container modules on top of each other and pack them with more steel spheres.

The startup has run a refrigerator-size pilot-scale system for 15,000 hours so far, and has built two much larger demonstration systems, one for a glass manufacturer and another for a cement producer, both in France. Those systems should be switched on in May.

“Being hybrid allows them to derisk the future,” Ceballos said. “Given the geopolitical problems today, it’s very attractive.”

In Brief

Bluesky’s website and app are still struggling on Thursday after experiencing service interruptions that COO Rose Wang attributed to a denial-of-service attack.

According to the social network’s status page, the issues began around 2:42 a.m. ET time on Thursday and have continued since.

Image Credits:screenshot of Bluesky

Currently, the Bluesky site and app will load at times, slowly, and other times will throw error messages.

For instance, switching to a particular feed within the app displays a message that says “This feed is currently receiving high traffic and is temporarily unavailable. Please try again later. Message from server: Rate Limit Exceeded.”

Image Credits:screenshot of Bluesky

Popular feeds like Discover or the official Bluesky Team’s feed are seeing this problem, though users’ own personal feeds may launch.

Other times, like when trying to visit a user’s profile, the site will just display an error message, forcing you to refresh and try again.

Image Credits:screenshot of Bluesky

Bluesky protocol engineer Bryan Newbold remarked around 3:46 a.m. ET, “oof, our services are getting pretty hard tonight.”

Reached for comment, Bluesky only pointed us to the status.bsky.app page and account (@status.bsky.app) for updates. The company did not comment on the cause or an ETA for a fix.

The service disruptions are impacting Bluesky, but other communities running their own infrastructure on the underlying protocol that powers the decentralized social network appear to be functioning for the time being.

Screenshot Image Credits:screenshot of Bluesky

Issues have continued into the afternoon on Thursday, as Bluesky’s status page says ” investigating an incident with service in one of our reginos [sic]” (That’s their typo, not ours).