An ex-Anthropic researcher’s doomsday warning comes at a very interesting time

An Anthropic researcher resigned this week, warning in a post on X that the company is “racing straight to self-improving superintelligence and gambling with our lives”. The company's own alignment lead even co-signed the message rather than walking it back. It's the kind of doomer warning the AI industry has flirted with before, but the timing, with Anthropic reportedly preparing for an IPO, makes it land differently.    On this episode of TechCrunch's Equity podcast, hosts Kirsten Korosec, Anthony Ha, and Sean O'Kane dig into the latest AI safety warning and what it says about the industry's race toward increasingly capable models. Plus, Apple's first big event under new CEO John Ternus and more of the week's headlines.    Listen to the full episode to hear more about:  Why Apple leaned hard into AI at its hardware event, and how always-on features baked into the Watch and AirPods may matter more than the splashy new foldable iPhone  How Stokes Space's $1 billion raise reflects a launch market that's suddenly wide open now that SpaceX is winding down its Falcon rockets  Why investors poured $2 billion into coding startup Cognition at a $48 billion valuation, and what it says about how much room VCs think is left in AI coding  What a $50 million seed round for floating nuclear reactors says about the scramble to power AI data centers  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.    Chapters:  00:00 Intro  3:28 Anthropic researcher's resignation warning  9:33 What's actually being done about AI risk?  15:56 Apple's AI-first hardware event  19:56 The case against a standalone AI wearable  21:53 Always-on listening: watch, AirPods, and privacy tradeoffs  26:43 The foldable iPhone's odds  27:16 Stoke Space raises $1B for a reusable rocket  31:17 Cognition raises $2B, betting on room for more AI coding players  34:13 BlueCore Energy's floating nuclear reactors  37:38 Outro  Learn more about your ad choices. Visit megaphone.fm/adchoices

AI companies have been talking about superintelligent AI like it's inevitable, but recent safety incidents like OpenAI's Hugging Face breach are demonstrating the potential dangers of deploying AI systems that are more capable than humans. So what happens when we can't reliably control what these systems do?  On this episode of TechCrunch's Equity podcast, Rebecca Bellan is joined by Connor Leahy, an AI researcher, entrepreneur, and now the U.S. Executive Director of nonprofit ControlAI, which is pushing for a far more radical approach to AI safety: stopping companies from developing superintelligence altogether. Leahy explains why he thinks the risks have become too great to manage through alignment and containment alone, and how what sounded far-fetched six months ago is now being backed by a wave of new legislation.  Listen to the full episode to hear more about:  His take on the Sanders-Casar “Ban Superintelligence Act,” the parallel legislation in the U.K. that Control AI advised on, and why he thinks the U.S. bill may go further than necessary.  How he views frontier AI labs less as companies simply chasing returns and more as political actors, and what that means for the trillions of dollars being poured into data center buildouts.  Why Leahy thinks the real point of no return is when AI can build better AI, creating a potentially runaway cycle of self-improvement  The case for international “trust but verify” agreements, and why Leahy thinks it isn't actually in China's interest to build superintelligence.  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.  Learn more about your ad choices. Visit megaphone.fm/adchoices

It's officially the Ternus era at Apple.   Tim Cook stepped down as CEO this week, handing the company to former hardware chief John Ternus, whose first memo promised a “huge launch next week” — timing that puts Apple's next iPhone event on his desk before he's even settled in. Cook isn't going far, though: he's staying on as executive chairman, focused on the kind of policy relationships that recently turned something as small as a map label into a very public balancing act. All of which raises an obvious question: what does the Ternus era look like, and how much rope will shareholders give him to figure it out?  On this episode of TechCrunch's Equity podcast, hosts Kirsten Korosec and Sean O'Kane unpack what Ternus is walking into, why he may actually be better positioned to make progress on software than hardware in this new AI era, and more of the week's news.  Listen to the full episode to hear more about:  Why Nvidia keeps making moves that look less like a chipmaker and more like a company trying to own the entire AI stack, from its Hugging Face acquisition to its investment in MediaTek and deeper compute deals  A chaotic week for robotaxis: Tesla's Cybercab event, Waymo's expansion into new cities, Zoox's first paid rides, and what happens when these companies start competing head-to-head  What Andreessen Horowitz's new $1.1 billion “Machine Age” fund signals about where early-stage AI hardware bets are headed  Why the $285 million acquisition of GoPro marks the end of an era for one of Silicon Valley's original consumer tech darlings  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.  Chapters: 00:00 Intro  1:12 Tim Cook steps down & John Ternus takes over as CEO  13:33 Nvidia's spreading its AI bets  21:58 A big week for robotaxis: Tesla, Waymo, Zoox & Uber  29:09 a16z's $1.1B "machine age" fund and $8.5B growth fund  32:39 GoPro's $285M acquisition and pivot to AI/defense  36:34 Outro  Learn more about your ad choices. Visit megaphone.fm/adchoices

The internet has a trust problem, and it’s not just because social media feeds are filling up with AI slop. AI-generated text and images are now making their way into job applications, product reviews, and even insurance claims, leaving platforms and users alike scrambling to figure out what’s real.  A handful of startups have cropped up in the past couple of years to become the “trust layer” the internet needs — including Pangram. The startup recently snapped up $9 million for its AI detection system and landed a partnership with Substack, which is now using Pangram’s tech to show readers which of their favorite authors use AI to write their newsletters. Pangram also recently dropped a new AI image detection tool.  On this episode of TechCrunch's Equity podcast, Pangram co-founder and CEO Max Spero joins Rebecca Bellan to dig into the promise of AI detection tools and where to draw the line between AI assisted and AI generated.  Listen to the full episode to hear more about:  Why Spero thinks the internet could be “dangerously close” to the dead internet theory becoming reality within a few years.  Why figuring out how much AI went into something is harder — and potentially more useful — than simply labeling it AI or human.  The stakes of getting AI detection wrong, especially when false positives involve sensitive images.  Why Spero thinks the “bottom tier” of writing jobs may be gone for good, while genuinely good human writing could become more valuable.  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod. Chapters:    00:00 Intro  1:15 Building Pangram: from binary detector to gauging AI assistance  2:06 Why job applications, reviews, and Amazon are drowning in slop  5:39 How Pangram actually spots AI writing  9:26 Where people draw the line on AI-assisted vs. AI-generated  12:50 The false positive problem, and "America has a Pangram problem"  14:30 Detecting fake images, now that AI photos are scarily good  19:13 Competing with GPTZero, Winston, and building "core infrastructure"  21:43 Will good writing jobs actually come back?  23:36 Outro   Learn more about your ad choices. Visit megaphone.fm/adchoices

Earlier this week, Meta agreed to pay $18 billion and make sweeping changes to how minors access its social networks to settle a lawsuit brought by 29 states. But the eye-popping settlement figure wasn't what caught the attention of the Equity podcast team. Instead, it was what Meta did next. The social media giant sent an open letter to TikTok and YouTube asking the companies to join it in setting industry-wide standards for teens, including daily time limits, blocking access to apps at night, and restricting notifications during school hours. But will these other companies follow? The Equity team doesn't think they will. On this episode of TechCrunch's Equity podcast, Rebecca Bellan, Kirsten Korosec, and Sean O'Kane dig into the Meta settlement, what it could mean for other social media companies, and more of the week’s headlines.  Listen to the full episode to hear more about:  How investors are pouring money into robotics startups and driving up valuations. For instance, Generalist reached a $3 billion valuation, while General Intuition hit the $6 billion mark. Even though robotics is a hot sector, there is pain point: robots still lack the know-how to do value-creating work. What senior reporter Sean O'Kane discovered when he dug into OSHA data connected to robotaxi companies like Waymo and Zoox. What self-driving truck startup Gatik's $200 million raise means for the sector. Hugging Face is reportedly being acquired by Nvidia. The team weighs in what this means for the industry. Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and wherever you get your podcasts. You also can follow Equity on X and Threads, at @EquityPod.  Learn more about your ad choices. Visit megaphone.fm/adchoices

Vibe-coding darling Lovable just raised $400 million at a $13.3 billion valuation, roughly doubling its worth in eight months. But Lovable isn’t the only Stockholm startup putting up huge numbers lately — legal AI company Legora and health tech startup Neko Health are right there with it.  On this episode of TechCrunch's Equity podcast, Dominic-Madori Davis is joined by Sophia Bendz, a partner at Cherry Ventures and longtime fixture of Europe’s startup scene, to unpack what’s driving startup growth in the Nordic ecosystem, from Sweden’s social safety net to the wave of U.S. capital flowing in.  Listen to the full episode to hear more about:  How the “Spotify Mafia” set the template for founders funneling their time, capital, and hard-won lessons back into Stockholm ecosystem, and how the same cycle may be starting again with the “Lovable Mafia”  Why Sweden’s social safety net might be fueling founders’ willingness to take bigger swings  Why Bendz is seeing American investors flying into Stockholm and handing out term sheets with little to no diligence, and the culture clash that’s creating with local funds  How Europe’s sovereignty conversation is shaping how founders and investors think about building, even as most of them still plan to IPO in New York  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.  Chapters:  00:00 Intro  0:50 Why Stockholm’s startup ecosystem is accelerating  2:34 How Stockholm’s “Spotify mafia” passes knowledge down  4:10 Founders House and building a culture of mentorship  5:50 Does Sweden’s social safety net make founders more ambitious?  10:00 Is there a “Lovable mafia” now?  10:00 Europe’s push for tech sovereignty  14:00 The U.S. venture capital flooding Stockholm  18:08 How Europe can build more growth-stage funds  19:45 Founders are asking VCs, “How can you help me?”  21:05 Stockholm’s tech conferences  25:51 Could robotics become Stockholm’s next big thing?  29:45 Outro Learn more about your ad choices. Visit megaphone.fm/adchoices

Andreessen Horowitz has two partners sitting on the boards of companies that now compete with each other: Ben Horowitz at Databricks and Martin Casado at Fivetran. Nothing too scandalous on the surface, except the Department of Justice has reportedly been investigating the arrangement for almost a year, dusting off a 112-year-old antitrust law that's rarely used against VCs.  Board conflicts aren't exactly new, and these companies weren't necessarily direct competitors when a16z first invested in them. But as portfolio companies expand into each other's markets, the DOJ's scrutiny raises a much bigger question for venture firms: How do you manage board seats when the boundaries between your portfolio companies keep moving?  On this episode of TechCrunch's Equity podcast, Kirsten Korosec, Anthony Ha, and Sean O'Kane dig into the a16z probe, what it could mean for VCs, and more of the week’s headlines.  Listen to the full episode to hear more about:  Why Stripe paid $7.5 billion for AI model router OpenRouter, and why the “singularity” isn't the real reason  What happens to the AI companies caught in the middle as OpenAI, Anthropic, and Nvidia pull further ahead  Why Rivian spinout Also just raised $150 million to make a bigger bet on autonomous vehicles  Uber’s newest delivery partnership with drone company Zipline, and what it means for the other autonomous startups betting their futures on Uber  Whether we've reached peak valuation for AI dictation apps after Wispr's $280 million raise at a $2 billion valuation  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.    Chapters:    00:00 Intro  1:02 TechCrunch Disrupt and the Startup Battlefield 200  3:17 The DOJ is investigating a16z over board seats  14:47 Why Stripe just paid $7.5B for “the Stripe for AI”  18:18 The mid-tier AI scramble: gateways, pivots, and acquihires  23:17 Anthropic's revenue surges while OpenAI's losses deepen  27:08 Also raises $150M and drops the “micro mobility” label  32:30 Uber teams up with Zipline for drone delivery  37:19 Wispr hits a $2B valuation  38:26 Outro  Learn more about your ad choices. Visit megaphone.fm/adchoices

The AI buildout shows no signs of slowing. And with hundreds of billions of dollars a year going into data centers and GPUs, compute has become the single biggest cost for anyone building AI products. But for all that spending, there still isn’t a straightforward way to put a price on compute — or for firms to hedge their exposure when the price changes.  Silicon Data just closed a $30 million Series A to change that. The startup aims to become the reference price for GPU rental and an index that a Wall Street futures contract would settle against. The company plans to launch its compute futures trading on the CME October 5th, pending regulatory approval.  On this episode of TechCrunch's Equity podcast, Rebecca Bellan is joined by Steve Hou, head of research at Silicon Data, to discuss the health of the AI buildout, and why the data is telling a different story than the doom and gloom headlines about depreciating chips and stalled data centers.  Listen to the full episode to hear more about:  What happens if OpenAI or Anthropic can’t make good on the massive compute commitments they’ve already promised, and whether “compute default” becomes a risk.  Why rental demand for older A100 chips is holding up, even as newer and more powerful GPUs hit the market.  If compute futures could make it easier for neo-clouds to finance expensive data centers by locking in future rental revenue.  What rising GPU rental rates across every contract length since March say about AI demand.  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.    Chapters:    00:00 Intro  0:59 Futures contracts explained  4:01 Can GPU compute actually be a commodity?  7:10 Who's really trading these: hyperscalers, AI labs, and market makers  14:32 Locking in prices to make data center debt easier to swallow  18:21 Inside the pricing data  20:25 The A100 mystery: why an old chip is still in high demand  22:56 Do GPUs depreciate as fast as everyone thinks?  23:20 Making sense of the Texas and New York data center pauses  25:40 What the futures curve says about where prices are headed  28:01 Silicon Data's $30M Series A and what's next  30:36 Outro  Learn more about your ad choices. Visit megaphone.fm/adchoices

Meta released Glimmer this week, an open-weight AI model anyone can download and run on their own hardware — a contrast to Muse Spark, the company’s more powerful model that stays locked behind its own APIs. The release landed alongside a letter from Mark Zuckerberg arguing AI should be “for everyone” rather than controlled by a handful of labs, but as Equity’s hosts point out, the vision comes with some asterisks.    On this episode of TechCrunch's Equity podcast, Kirsten Korosec, Anthony Ha, and Rebecca Bellan take a look at Glimmer, Zuckerberg’s 6,500-word manifesto, and more of the week’s headlines, from the true cost of the AI industry’s energy needs to a $250M acquisition gone very wrong.    Listen to the full episode to hear more about:    Who’s on the market for a $300 cocktail robot?  Why Anthropic is adding watermarks to its generated text, and users are not happy.  The potential cost Amazon’s planned data center in Texas and the startups racing to fix the grid, including Form Energy’s $750M raise for 100-hour batteries, Reservoir’s $8M bet on smarter water heaters, and Discovered Materials’ hunt for cooler chips.  Why Joby Aviation’s $500M acquisition of a defense contractor reminds Kirsten of its Blade deal last year — and what the 2028 LA Olympics have to do with it.  How a $250M deal between video-clipping startup VideoVerse and sports publisher Minute Media collapsed amidst allegedly forged documents, multiple lawsuits, and a CEO nobody can reach.  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.    00:00 Intro  1:20 Would you buy a $300 cocktail robot?  3:08 Meta's Glimmer and the “AI for everyone” pitch  8:03 Anthropic starts watermarking its writing, and users have thoughts  14:34 Amazon's data center and its natural gas problem  16:37 The startups racing to keep the grid from melting down: Form Energy, Reservoir, and Discovered Materials  23:28 Joby Aviation buys into defense with Resonant Sciences  28:24 The $250M VideoVerse-Minute Media deal, a missing CEO, and a mess of lawsuits  32:37 Outro  Learn more about your ad choices. Visit megaphone.fm/adchoices

AI notetaking hardware has taken off over the past couple of years, with credit-card-sized devices, pendants, pins, and even transcribing earbuds all promising to capture your meetings and turn them into summaries and action items. Now, a whole wave of wearables — rings especially — are betting people want to capture stray thoughts and ideas the same way.    One of the companies chasing that bet is Sandbar, the startup behind the private voice ring Stream, which has raised $36 million to date, including a $23 million Series A led by Adjacent and Kindred Ventures.    On this episode of TechCrunch's Equity podcast, Rebecca Bellan talks with Sandbar co-founder and CEO Mina Fahmi about why he thinks so many voice hardware devices before Stream struggled to break through, and why he's betting that keeping the human firmly in control is what it'll take to get wearable tech right.    Listen to the full episode to hear more about:  Why Fahmi designed Stream to be push-to-talk instead of always-on, and what he thinks “social acceptability” means for wearable AI.  How Fahmi's time at CTRL-labs, the neural interface startup Meta acquired in 2019, shaped his thinking on human-driven input.  Why Sandbar chose the “gruelingly painful” path of building fully custom hardware instead of reskinning an off-the-shelf device.  What Fahmi thinks it'll take for voice wearables to break out of tech-enthusiast circles and into the mainstream.  Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.  Learn more about your ad choices. Visit megaphone.fm/adchoices