Techin' Geek

If there’s any doubt about how seriously Meta is taking AI, here’s a new signal: The company on Wednesday announced that its Muse assistant is now available in a dedicated iPad app. That’s a milestone it took Instagram some 15 years to reach, despite constant user demand.

Muse, meanwhile, hit the iPad just a month after launching on iPhone.

(To be fair, it may not be as difficult to bring something like Muse to iPad, given its text-heavy interface, where image and video resolution and formatting matter less.)

Since launching on iOS and Android on September 8, Muse has topped 6.6 million installs, according to estimates from market intelligence firm Sensor Tower. The app, which is now one of dozens of consumer-facing AI agents, lets users connect their accounts to stay on top of email, meetings, bills, and more, and to complete tasks like booking reservations, ordering groceries, setting goals, and making purchases.

To make that work, Meta has been adding partners in the effort, which appear in Muse as “connectors.” Users can add and sign in to them selectively, depending on which parts of their online life they’re comfortable letting an AI agent handle.

With the iPad release, Meta has updated this list with several of its newer additions, including connectors aimed at small businesses, like Asana, Canva, Dropbox, Figma, QuickBooks, GitHub, Klaviyo, and Zoom, as well as connectors for Meta ad accounts. It also added Notion and Granola.

In addition, Meta recently announced more retail partners, like Best Buy, Gap, Sephora, Walmart, and Wayfair. These allow the agent to shop those retailers’ sites and complete transactions on users’ behalf. Because this technology is still new, these experiences can still run into issues as websites mistake agents for bots and block their access.

On Tuesday, Meta announced that it’s working with industry partners on a technological solution to this problem in the form of an open standard that would allow personal AI agents to identify themselves to businesses. That would help businesses sort the good bots from the bad.

A now-former officer with the Central Intelligence Agency has pleaded guilty to stealing almost $200 million, including hundreds of gold bars, as part of a fraudulent scheme masquerading as a top secret government program.

The U.S. Department of Justice on Tuesday confirmed that David J. Rush, a Virginia resident, admitted to one count of wire fraud linked to running the fabricated operation and defrauding the government while he was a CIA officer. In a statement, prosecutors said that Rush lied about his education and military experience to gain a position at the CIA, and used his access as an employee to set up the top secret and highly compartmentalized program with the intention of using it to funnel U.S. government funds into his possession.

During his arrest, law enforcement found more than $46 million worth of gold bars in Rush’s house linked to the long-running fraud scheme, along with cars, watches, and over $2 million in cash.

The plea deal means that Rush escapes trial, as much as the U.S. government likely avoids having to explain in open court how Rush evaded vetting scrutiny to join the U.S. intelligence agency, and how actions went undetected for more than a year.

Rush was until recently a senior CIA officer, who according to earlier reports worked for a division of the CIA since 2010 that develops hacking tools and techniques for carrying out espionage. Per court filings, Rush had significant authority over government intelligence programs and spending.

Over the course of 2025, Rush used his position to fabricate a so-called special access program, which are surveillance or intelligence-gathering operations so sensitive that only a few people are read into them. Special access programs, or SAPs, require special clearance that not even the most cleared individuals in government are granted access to. According to The Washington Post, Rush constructed the compartmentalized program under the guise of a fake “continuity of government” plan, aimed at keeping the government functioning in the event of a war, natural disaster, or some other destructive catastrophe. As part of the scheme, Rush used the program and fake government contract to trick an unnamed defense contractor into buying large amounts of gold that Rush then pocketed.

According to The Post, the case astounded officials as Rush’s duties at the CIA involved work with one of the government’s “most sensitive intelligence-gathering programs,” with details so sensitive that only a few senior U.S. intelligence officials and lawmakers know of its existence. The Post withheld details of that program from publication as its disclosure could “jeopardize ongoing intelligence-gathering operations.”

Court filings said that by virtue of his position, Rush “effectively acted as his own approving official, enabling him to expend substantial amounts of United States government money without meaningful scrutiny.”

Rush is scheduled to be sentenced in late January 2027 and faces up to 20 years in prison.

When self-taught coder Sigil Wen was 17, he moved to Silicon Valley and lived in an AI hacker house with famed AI researcher Andrej Karpathy.

While there, he hacked and coded alongside other people who would become the biggest names in AI, like Perplexity founder Aravind Srinivas and OpenAI researcher Noam Brown. He tested early versions of AI tools that would later become well known, including a chatbot shared by Anthropic co-founder Ben Mann that would become Claude, an image generator from David Holz that would become Midjourney, and what would become OpenAI’s GPT-3 and the image generator Stable Diffusion. Prominent investor and entrepreneur Naval Ravikant hired him for Airchat, Ravikant’s now-defunct rival to the Clubhouse social network.

For fun, he figured out how to get GPT-2 running on his Apple Watch. “It was a magical time,” Wen told TechCrunch.

Now a Thiel Fellow — the program from investor Peter Thiel that invites young founders to pursue projects instead of college — Wen on Monday launched an invite-only beta of Underdog, one of the most private AI assistants Silicon Valley has yet to offer. The model runs wholly on-device, meaning the user’s data remains on devices they already own, currently Macs and Windows PCs, with Linux, iPhone, and Android versions coming soon.

Underdog is powered by Husky, the inference engine Wen built to run AI models quickly on a user’s own hardware. Wen says it moves less data between a computer’s main chip and its graphics chip than rival on-device engines.

Underdog has other security features baked in, too, like encrypting the keys to the email and other accounts that users authorize Underdog to access.

Underdog is, however, using much smaller models than today’s state-of-the-art ones hosted in data centers. It currently uses a 27-billion parameter reasoning model fine-tuned from Qwen3.8-27B.

Wen argues that this model compares favorably with Claude Opus 4.6 in some benchmarks, or what was considered top performance six months ago. He says that means it can handle the everyday tasks that people want an AI assistant to do, like shopping research or answering math homework questions.

“You don’t need to sacrifice your privacy for the capability because they’re just as capable,” Wen says. He adds that small on-device models will continue to grow more capable over time.

Perhaps the most interesting thing about Underdog is its early business model. The app will be free at first and never ad-supported. Since the AI runs on users’ machines, Underdog doesn’t have the giant overhead of paying a provider for inference. “I don’t have to charge you a subscription to run this because my costs are so super low,” he said.

Instead, with Stripe co-founder Patrick Collison as one of his angel investors, he’s borrowing a play from the fintech era. Underdog will take a tiny percentage of payment transactions that the AI assistant makes using Stripe’s secure payment rails, something like an interchange fee. In this way, the AI assistant never has to mine your data. It is as aligned with you as your bank or credit card providers.

This runs contrary to the business motivations of many of the other players in AI assistants, whose privacy policies allow them to collect data on users that they may sell to advertisers or other third parties, or use to train other models.

That kind of data collection could be a particularly treacherous trade-off for users of an AI assistant, which may need access to the most intimate details about you in order to be useful, from medical conditions to financial data to data on your kids.

As Wen wrote in what he calls his AI manifesto, “Why should using AI require surrendering your private information?”

He tells TechCrunch: “I honestly want to build Underdog for myself. I’m building a product that I would be proud for my future children to use.”

The startup behind Underdog is named Conway Research, and Collison isn’t its only big-name investor. Conway is backed by Andreessen Horowitz via partner Chris Dixon, as well as Khosla Ventures, Hummingbird, SV Angel, the Anthology Fund (the partnership fund between Menlo Ventures and Anthropic), and a prominent list of angel investors that includes Vercel founder Guillermo Rauch, Noam Brown, and Deedy Das, among others.

When Flai was raising its seed round last year, it was just a team of three people pounding the pavement to get car dealerships to use the startup’s software to manage phone calls, emails, and texts. But CEO Ari Polakof was already looking to a more agentic future, where Flai’s AI would handle far more.

One year later, that future has arrived. The startup is working with more than 10 of the top 50 dealer groups in the country, and its AI is being used to answer and engage with customers, run outbound campaigns, and schedule appointments — 50,000 per month — across both sales and service. This has led to a 20x increase in revenue, Polakof told TechCrunch in an exclusive interview.

Flai says this deep integration is helping the startup set itself apart from others trying to get car dealers hooked on AI. The growth is attracting investors, too. On Tuesday, Flai said it had raised a $27 million Series A funding round led by automation-focused firm Base10 Partners. The round included funding from dealers (Friedkin Group and Findlay Automotive), Toyota’s venture arm, Y Combinator, and First Round Capital.

Polakof said Flai’s growth is coming from customers who see the bigger vision of what is essentially an AI-powered customer relationship management software built from the ground up and tailored to the specific need of each dealer.

“Sure, we do answer the phone calls, but we’re following up with customers, we’re alerting the leadership whenever they need to take a look at something, maybe because the customer’s mad, or because you forgot to respond, or promised them something next week,” Polakof said. “We’ve grown quite a bit from just a phone solution to more of a platform. There are plenty of AI companies coming in, but I think that really speaks about the opportunity.”

Flai claims its software is also flexible. Earlier this year it launched with a luxury dealer in Puerto Rico where customers “switch between Spanish and English mid-conversation,” Flai co-founder Juan Alzugaray wrote in a LinkedIn post. “We started where we always do: inbound service calls. It went well enough that they added sales. And now the group is rolling Flai out across all 8 of their stores,” he wrote.

The result of all this is that dealerships are selling more cars, making more money, and “customers are getting answers faster,” according to one of the groups Flai works with.

Perhaps unsurprisingly, Polakof said much of Flai’s customer growth has come from the dealers talking up the product to their peers.

“If you really do great by them, they also do great by you, and they recommend you, and talk about you,” he said. This has led to Flai generating half of its revenue from new customers, according to Polakof.

Customers are making these referrals because they like Flai’s software, Polakof said, but also because they get up and running quickly. Flai’s software can be live in a dealership 10 days after a contract is signed, and he tells his team — which is now up to around 40 people — that they need to respond to customers within 20 minutes “at most.”

“The common experience, especially around automotive [software] vendors, is — and many are almost traumatized by this — is that they sign the contract and then the company disappears,” he said “I’m fully focused on making sure that we treat every customer as if they’re our only customer, especially because I know how hard it was to get the first one.”

That speed and customer experience, combined with the knowledge Flai has developed on how dealerships work and what they need, is what Polakof said protects the startup from products like Muse, which Meta has pitched as a one-shot solution for small businesses.

“You get a group of really smart people, and you spend all your days, all your time thinking very deeply about this one space — it’s just so hard to envision that any other platform outside of the space can come in and start going after your customers,” he said.

For all this momentum, there is still one customer Polakof hasn’t been able to land: his own local dealership.

“I’m actually trying much harder than I should to make them a customer,” he laughed. “I’m sure we will get them by next year.”

Lucid Motors built 2,954 electric vehicles (EVs) in the third quarter of this year, a 54% drop from a year ago, as the company purposely limits production to better meet demand for its EVs.

This was the third straight quarter in which the number of EVs Lucid built has declined. It’s also the lowest quarterly output since the first quarter of 2025, which was just after Lucid Motors started production of its second EV, the Gravity SUV.

Lucid delivered 3,806 EVs in the third quarter, roughly flat with the second quarter and down about 200 vehicles from the third quarter of 2025. The company has struggled to find buyers for either of its first two luxury EVs. In five of the last six quarters, it built more vehicles than it delivered.

Lucid’s new CEO, Silvio Napoli, has spent the last few months leading an effort to “simplify the company.” That effort has included laying off around 1,500 employees, streamlining the company’s leadership, and eliminating a second shift at its factory in Arizona in a bid to reach cost savings of $1.4 billion. Lucid also delayed the release of its third EV, the Cosmos. That model is supposed to be much cheaper, starting at under $50,000.

The third quarter figures, released Monday afternoon, come just a few days after rival EV upstart Rivian posted its best quarter in history on the back of the R2, its new, more affordable SUV. Although Rivian didn’t break out specific delivery figures for the R2, the company shipped nearly 20,000 vehicles in the third quarter, the first full quarter with the R2 in production, up from 12,194 in the second quarter.

Lucid’s failure to find a large market of buyers for its EVs is even more stark when compared with the promises the company made when it went public in 2021. That year, Lucid Motors merged with a special purpose acquisition company and estimated it would ship as many as 90,000 EVs in 2024 alone. The company raised $4 billion in the transaction.

On Lucid’s second-quarter earnings call in August, Napoli spoke about why he thinks the company has failed to make a dent in the EV market.

“While there is no question that Lucid brought leading innovations and outstanding products to the market, we have disappointed on several fronts, and for far too long,” he said. “We have not executed consistently. We missed commitments, launched products before they were ready, underinvested in service, responded too slowly to quality issues, and allowed complexity to slow decisions down.”

The Cosmos’ lower price could, in theory, let Lucid access a wider market, but Napoli cautioned shareholders that rushing the new EV out could create more trouble.

“We will not repeat the mistakes of the past by bringing a product to market before it is ready,” Napoli said on the call.

OpenAI will start adding an invisible watermark to text generated by ChatGPT and Codex in the European Union to comply with the EU AI Act, the company said Monday in a blog post.

The EU AI Act’s transparency rules, which took effect on August 2, require AI companies to mark AI-generated content in a way other systems can identify.

OpenAI said the watermark will roll out over the coming weeks to eligible ChatGPT and Codex users on all plans, but only in the EU. Developers using OpenAI’s API anywhere in the world can turn it on for select models starting today; it’s off by default. OpenAI said it is not making text watermarking a global default at launch.

The watermark is not an actual symbol, but works by subtly shaping the model’s word choices, leaving a pattern readers can’t see, but a detector can pick up. Because it lives in the words themselves, it travels with the text when it’s copied and pasted. OpenAI said the watermark doesn’t identify the user, and that it saw no meaningful change in its models’ performance with it switched on.

OpenAI also published a technical report for its method, called textGrain, alongside the announcement. Co-written with researchers from the University of Pennsylvania and Yale, it walks through an example of using a secret key to sort next-word predictions to finish the sentence. Add hundreds of these nudges together, and the detector can spot AI-generated content using only the text and the key.

Can the watermark be removed by editing? OpenAI’s tests suggest yes. In one test, replacing 10% of words with synonyms dropped detection from about 92% to 66%. The company also said short passages, math answers, and translated text are harder to detect.

“These limitations contribute to our decision to provide initial detector access only to approved researchers and expert organizations, who can help us evaluate reliability and responsible uses,” said the company.

OpenAI also cautioned that a missing watermark “does not prove human authorship.” The text could be too short or too heavily edited, or it could come from another company’s AI.

“[Watermarks] can indicate that an OpenAI system generated or processed part of a passage, but not how much human judgment, editing, or creativity went into it,” the company said.

The announcement comes two months after Anthropic said it would watermark text generated by Claude, a move it’s applying worldwide. That decision drew backlash from some Claude users, who argued they had supplied “the instructions, context, decisions” while Claude was just “the tool.”

OpenAI had built a text watermark before but held off on releasing it, partly over concerns that users would switch to rivals that didn’t watermark, The Wall Street Journal reported in 2024.

Anthropic, Google, Meta, Microsoft, and OpenAI are among the companies that have committed to following the EU’s code of practice on AI-generated content.

Although it’s only been a couple of months since Etched raised $700 million at a $21 billion valuation, the AI chip startup is already being plied with investment offers at double or more its value, according to people familiar with the company.

Etched is reviewing incoming bids that range from $40 billion from top-tier investors to $50 billion from lesser-known backers, one person said. These fundraising talks are early, so terms of any deal, should one happen, may change. Etched declined to comment.

While this may seem like a fast timetable to raise another mega round, Etched is pursuing a particularly expensive segment of the AI industry: building full AI hardware systems powered by its own proprietary chips. The person familiar with these offers said that if it raises as much as its last round, this could give Etched a cushion of as much as 3.5 years of runway.

There are reasons why VCs are hot to own a piece of Etched. The four-year-old startup shows promise of challenging Nvidia. Not only did quant trading firm Jane Street lead the last $700 million round, it is also a customer that took delivery of an early system. Etched said in July that it had already secured $1 billion in customer orders, including the one from Jane Street, after manufacturing its test chip at a TSMC factory this summer.

Co-founder and COO Robert Wachen previously told TechCrunch that investors are so enthusiastic because Etched has designed two new components from scratch to speed up inference — the computing process that happens after a user submits a prompt.  

The company claims its chips can process more tokens faster and at a lower cost than Nvidia’s. That’s the reason its processors have been so attractive to Jane Street, for whom a microscopic advantage in speed can yield massive profits.

The startup has also impressed investors with its ability to attract engineers from Nvidia, with roughly 15% of Etched’s 400-person workforce having previously worked at the chip giant, according to The Wall Street Journal.

Etched also operates a new 10-megawatt data center in Silicon Valley and established a facility in Taiwan to coordinate production near TSMC.

Co-founders Gavin Uberti and Chris Zhu famously met in an advanced math course at Harvard, while Wachen was Uberti’s roommate and they dropped out to pursue the company.

In terms of fast rounds at big leaps in valuations, Etched already has a history of them. The startup announced a $300 million round at a $10.3 billion valuation led by Sequoia in July. It announced the $700 million round at a $21 billion valuation in September. Back-to-back funding rounds, which essentially act as a single financing split into two tranches with separate valuations, are increasingly common among the buzziest startups.

Menlo Ventures partner Matt Murphy and his team announced Monday in a blog post that the firm has invested in AI coding startup Factory.

Menlo declined to comment on how much it invested but said the deal was part of Factory’s latest funding round announced last month at a $5 billion valuation. A source familiar with the deal told us the investment was significant to Menlo — not a token gesture — and that Menlo would have invested more had there been room on the cap table.

Ordinarily, something like this wouldn’t necessarily be news; startups often extend previous rounds to add new investors. But last week, Factory co-founder and CEO Matan Grinberg was embroiled in a very public airing of dirty laundry when he alleged that he had fired investor Chris Degnan of firm RPT Partners from his role as board advisor. Grinberg said he feared Degnan may have shared confidential information with Factory’s biggest competitor, Cognition. The accusation followed Degnan’s move to Cognition as its chief revenue officer.

A who’s who of the tech industry came out of the woodwork either to support Degnan and condemn Factory, or vice versa, but none was more surprising than Vinod Khosla. Khosla’s firm is an investor in both Factory and Cognition. But that didn’t stop the venerable VC from calling Factory a desperate “struggling second tier competitor,” and accusing Grinberg of lying about firing Degnan and impugning his character. Degnan disputed Grinberg’s story, saying he resigned and that he’d rebuffed a competing job offer from Grinberg.

So Menlo’s full-throated endorsement of Factory, which praised its founders, its tech, and its relationship with its customers, is more than a feather in Factory’s cap. It’s a statement that Factory is not in the state that Khosla implied it was.

Murphy famously bet his firm on Anthropic back when it appeared to be an also-ran to OpenAI and has been on a hot streak ever since, landing deals with Lovable and Legora, for instance. So now some VCs are calling Factory “the next Anthropic.“

Sequoia’s Shaun Maguire, also a Factory backer who came out on X in support of Grinberg last week, appears thrilled with Menlo’s endorsement, saying the firm is “on a tear.” Notably, Menlo is not an investor in competitor Cognition.

President Donald Trump hosted many of the biggest names in artificial intelligence this week — in part to announce that the U.S. government isn’t calling it artificial intelligence anymore. Now it’s super intelligence.

On the latest episode of TechCrunch’s Equity podcast, Kirsten Korosec, Sean O’Kane, and I discussed the motivation behind the meeting and what effect it might have on the AI industry. We were all pretty skeptical that the “Joint Commitment on Frontier Responsibilities” that the executives signed will amount to much, especially since — as Sean noted — it’s “deeply non-binding,” at least from a legal prospective. (Instead, Trump said he considers it “morally binding.”)

Kirsten suggested that the meeting’s real significance may have been as “a rebranding effort around AI.” (An effort that continued this weekend with Trump’s announcement of a new Super Intelligence Force.) For her, the big message was, “AI is going to kill us, it’s scary, it is going to take jobs — but super intelligence is not.”

Keep reading for a preview of our conversation, edited for length and clarity.

Anthony Ha: Officially, by order of the president of the United States, it’s not artificial intelligence anymore, it’s super intelligence. We were talking about this last week, because this is something [Trump has] been hinting at — he doesn’t do hints, he’s been saying [it] for a while. But now it’s signed into executive order. If you’re an official U.S. representative, you are supposed to say super intelligence, not artificial intelligence.

And this was also part of this broader announcement around AI safety, which I think is both interesting for this pact that was announced, but also the fact that it happened at all. It seemed to have come together fairly quickly, because just two weeks ago, a week ago, President Trump was saying, “We don’t need any of this, we need to lead in AI, we can’t constrain it in any way, we don’t need any regulation, anyone who’s saying that there’s any danger from AI is essentially a Chinese or Democratic plant.” And yet all these AI CEOs got together with him earlier this week and signed this pact.

Kirsten Korosec: Leading up to that, we should say, not only was there pushback on regulation, and we can talk about whether this executive order even does any of that — I would say no, there’s a little bit of theater there — but the important and interesting thing is that just days ago, Anthropic was really in the doghouse, if you will, in terms of what was happening behind the scenes with President Trump. 

That all changed when [Anthropic CEO] Dario Amodei was invited to a 10 p.m. dinner. He had this sit-down, and then just days later, we had this luncheon, and at this luncheon, we have pretty much every tech power broker from every major tech company that exists. So Mark Zuckerberg was there, Jeff Bezos was there, Elon Musk was there, Dario of course was there, they were all there. 

I’m wondering if any of you watched the press conference that happened afterwards, which there’s been many, many memes about, because of the interesting, funny behavior when you bring a bunch of tech CEOs together.

But I guess my bigger question for both of you is: What does this all mean? And does it mean anything?

Sean O’Kane: How much time do you have? First off, sure, they had dinner, Trump and Dario. If you read the Time magazine interview that Trump just did, your guess is as good as mine as to how informed he was that that dinner was even going to be happening, which is a strange reality to live in, but it’s just the one we inhabit.

I will also say one other thing that makes me skeptical — surprise, surprise — of how repaired the relationship is between the administration and Anthropic, is that for as much as there are clearly people trying to put those two groups together, the Defense Department is still very adamant that Anthropic is terrible and woke and and a national security risk. Emil Michael goes off on Anthropic every other day on X, it seems. I think there’s still some distance there despite Dario showing up to this thing. 

As far as the agreement and all of that, let’s be real, they misspelled the United States on the agreement that they all signed. It is also just deeply non-binding. I mean, it is completely voluntary. I think my favorite aspect about it was that Trump described it as quote “morally binding,” which is just a level of abstraction and terminology that I’ve never heard of before.

Anthony: In terms of Anthropic and their relationship with the Trump administration, there’s been a sort of a lot of reporting, especially during that big blow up with the Defense Department, that different parts of the White House had different relationships with Anthropic. The Treasury Secretary seemed to be a big fan, whereas the Defense Department — definitely not big fans, or at least unhappy with some of the conditions that Anthropic were trying to put on the use of their technology. 

In  that sense, it feels like nothing has necessarily changed. There are people in the White House who want to use Anthropic’s technology, want them involved in these decisions, and others who really, really, really, really don’t.

In terms of the agreement, it seems not super meaningful. What is meaningful is just the fact that they felt that this was something worth doing. As indicated by the fact that they misspelled United States, it seemed like they rushed to put all of this together. [But] the fact that this kind of gesture was needed seems significant, even if I don’t think the agreement itself is going to lead to any major changes.

Kirsten: I do think that there was one important thing that happened. And to me, this is the point of it, which was a rebranding exercise around AI. “AI is going to kill us, it’s scary, it is going to take jobs, but super intelligence is not.” This isn’t my personal belief. To me, that was the message.