Tech Brew Ride Home - Anthropic Doesn't Hate Open Weights, Says Anthropic.

Episode Date: July 28, 2026

Dario Amodei said Anthropic never backed an open-weights ban, pitching mandatory safety tests instead as OpenAI and Google signed on. Altman headed to Washington, Korea's KOSPI cratered 11% on AI jitt...ers, Apple launched Klarna leasing, and shipped 194 CVE fixes. Anthropic wants tests, not bans, as OpenAI and Google back open weights (The New Stack) Source: Sam Altman will meet with senior US officials, lawmakers, and economists in Washington, DC, this week to preview OpenAI's upcoming family of AI models (CNBC) South Korea's KOSPI drops 11%+, led by chip stocks, amid concerns over China's chipmaking progress and the AI spending boom; Samsung falls 11%+ and SK Hynix 12% (Bloomberg) Credit default swap prices tied to Oracle, SpaceX, Alphabet, Amazon, Meta, Broadcom, and Nvidia hit record highs as investors turn jittery over Big Tech's data center debt; Oracle's five-year CDS reached 215bps (FT) Apple launches Apple Upgrade, a new US leasing program in partnership with Klarna that replaces the iPhone Upgrade Program, starting at $17.99/month for iPhones (MacRumors) Apple releases 26.6 updates for iOS, macOS, iPadOS, watchOS, tvOS, and visionOS with a huge number of security fixes; macOS Tahoe 26.6 alone addresses 155 CVEs (9to5Mac) Subscribe to the ad-free feed. Learn more about your ad choices. Visit megaphone.fm/adchoices

Transcript
Discussion (0)
Starting point is 00:00:00 Now streaming on Disney Plus, the Hulu original series Furious follows FBI agent Alice Black, played by Emmy Rossum, on the hunt for a mysterious and calculating serial killer. Both walk their own paths toward justice, and as their lives start to intertwine, the line between right and wrong begins to blur. New episodes week over week, watch the Hulu original series Furious, now streaming only on Hulu on Disney Plus. Welcome to the TechBrew right home for Tuesday, July 28th, 2026. I'm Brian McCullough today. Dario Amadai said Anthropic never backed an openweight's ban, as others have been signing openweight AI acceleration letters. Sam Altman headed to Washington, Korea's Copsie, cratered on AI jitters, Apple launched its leasing program, and yo, if the software update knocks these few weeks coming up, you better answer the door. Here's what you miss today in the world of tech.
Starting point is 00:01:01 So I thought all along that this was kind of inside baseball, so I haven't told you about the online chatter about how people have been accusing Anthropic of being anti-openweight AI, accusing them of wanting regulatory capture by asking to be regulated. But, well, Dario Amma Dye wants you to know Anthropic has never backed, say, an open-weight model ban, quoting the new stack. Anthropic does not want open-weight AI models banned. CEO Dario Amadai said so plainly today after a week of criticism. Instead, he proposes mandatory safety testing for every sufficiently capable model before release. The catch is that nobody has defined where sufficiently capable begins, and that determines whether tomorrow's next open-weight models stay available. Anthropic has never advocated for a ban on open-weight models. Amadai writes in his letter, calling models without dangerous capability. a public good that costs nothing beyond the compute to run them. His proposal has three parts. Titan export controls on advanced chips and chipmaking equipment, cracked down on industrial scale model distillation, and require safety evaluations for every sufficiently capable model before release. The first two are aimed at Washington. The third would affect developers preparing models for release. Amadai calls the testing idea close to consensus, noting the Trump administration has moved that way and that recent industry proposals would apply testing regardless
Starting point is 00:02:34 of a model's origin or license, while, quote, exempting less capable models such as those from startups and academia entirely. The exemption is critical, though. Again, the term startup can be hard to pinpoint. Once some models are exempt and others are not, the policy becomes a capability threshold by another name. So where does the line fall? Well, no one has said. The letter does not name who would set the number or who would run the test. Model capability could be measured by metrics like benchmark results, training compute, or scores evaluated against defined risk categories, and the chosen metric would dictate the model type. Whoever picks the metric is also setting release timing for every lab that publishes weights. This is a fast-moving idea. The OpenWates
Starting point is 00:03:21 and American AI leadership letter so far counts 50 signatories with OpenAI and Google joining Microsoft Nvidia, META, AMD, Cisco, Cloudflare, GitHub, and Olamma, Anthropic, and Amazon, remain notable holdouts. On Monday, NVIDIA launched the Open Source AI Alliance with more than 35 founding members, including Microsoft, IBM, Hugging Face, Palantir, Red Hat, Crowdstrike, and the Linux Foundation, Anthropic is not among them. Invidia's argument is that defenders need open frontier models to investigate incidents, citing the recent case in which closed tooling blocked forensic work on a hugging face incident and Hugging Face ran an analx on open weight models instead. To his credit, Amadai addressed the
Starting point is 00:04:07 obvious criticism directly. A U.S. ban, he said, would mostly shield domestic AI companies from competition while doing little to stop bad actors. It would protect U.S. AI companies from competition, but that has never been my goal, he said, end quote. That's fair. It's also true that Anthropic does not release open weight models, so the statement feels a touch empty. Amadai concedes the testing, quote, would need to be global, which means even the CCP would need to be on board and figures that may be possible. This could be the largest unanswered question around the story. Expert controls and anti-distillation measures are policies Washington can oppose on its own. A global capability testing regime is different. The proposal that most directly affects open
Starting point is 00:04:51 weight models depends on broad international agreements that do not exist yet and may never exist, end quote. We're going to come back to this soon. OpenAI CEO Sam Altman will meet with senior Trump administration officials, lawmakers, and economists in Washington, D.C. this week to preview the capabilities of the company's upcoming family of artificial intelligence models CNBC has learned, quoting them. Altman has maintained a regular presence on Capitol Hill this year, as he's tried to alleviate policymakers' concerns about the rapidly advancing technology and help shape their views about what regulation should look like. This visit lands during a particularly sensitive moment for AI development in the U.S.
Starting point is 00:05:38 as a debate over whether to restrict Chinese open-weight AI models has ripped through Silicon Valley. In addition to previewing Open AI's upcoming releases, Altman is expected to field questions from policymakers about cybersecurity and the company's stance on open-weight models, according to a source familiar with the plans who asked not to be named because the details are confidential. Openweight AI models are available for users to download, modify, and run on their own infrastructure. In recent weeks, Chinese openweight models from startups like Moonshot AI have rapidly caught up with leading proprietary models from American companies, including OpenAI and its chief rival, anthropic, sparking anxiety about the durability of the U.S. lead in the AI race. officials and executives have been weighing whether or not to limit access to Chinese models in the U.S.,
Starting point is 00:06:26 which prompted a coalition of tech companies, including Nvidia, Microsoft, Meta, and Palantir to release the letter on Friday, urging policymakers to avoid premature restrictions, their words on Open AI models. OpenAI signed the letter after it was initially published, but the company's allies have lobbied for restrictions on open-weight models, according to a report from the New York Times. Anthropics still has not signed or made an official comment about its position, although see the previous segment. Nvidia CEO Jensen Huang is also meeting with lawmakers in Washington, D.C. this week to discuss open models and, quote, American leadership in AI, according to a spokesperson.
Starting point is 00:07:06 While in Washington, Altman is also expected to discuss the evolution of AI agents and their impact on worker productivity, the person said he will explain the concept of AI teams, where multiple agents work together to carry out long-term tasks, end quote. Another thing I don't know how to cover is basically the increasing ups and downs of the market reaction to the potential ups and downs of conventional wisdom of whether or not we are in an AI bubble, although, you know, half the fun of living through the dot-com era was that, the ups and downs, the horse race, even though the market seems to be up today, I'm still. Still noting that yesterday, South Korea's COPSY index dropped more than 11%.
Starting point is 00:07:57 That's an entire nation's stock market index, led by chip stocks amid concerns over China's chipmaking progress and the AI spending boom. Samsung fell more than 11%. S.K. Heenix, 12%, quoting Bloomberg. When one trade becomes this crowded investors don't wait for bad news, they simply need a reason to take profit, said Villaletta Torterova, a senior research analyst, at leverage shares. After an extraordinary rally, expectations have become almost flawless when valuations leave no room for disappointment. Even small changes in sentiment can trigger deep corrections, end quote. The news has raised further questions about whether hundreds of billions of dollars
Starting point is 00:08:36 being spent on AI development would be worth the payoff. Investors are also jittery about high valuations for global semiconductor stocks. Moreover, in Vidia's $750 billion in infrastructure deals have added concerns over AI-related debt levels. Fears of so-called circular deals are on the rise, as AI investments are increasingly becoming an interconnected web of dependencies between technology manufacturers and AI startups that can magnify losses if demand for AI fails to match lofty expectations. Doubts over spending returns and valuations are still deepening rather than fading,
Starting point is 00:09:12 said Heavy Chen, a senior marketing analyst for Vantage Global Prime. The hesitation to buy the dip suggests investors are waiting for stronger proof before rebuilding exposure, end quote. The losses underscore how quickly sentiment has turned on one of the market's most coveted trades. The costs of protecting Nvidia's debt against default surged by a record on Monday. That signals, quote, the credit market noticing something that equity market hasn't fully priced, said Dylan Wu, a research strategist at Pepperstone Group Limited, and video shares fell by 0.9% on Tuesday, end quote. So, yeah, on that specifically, quoting the FT,
Starting point is 00:09:52 a closely watched gauge of risk in the holding of the debt of companies at the center of the AI boom is rising rapidly, underscoring growing jitters over big tech's vast spending on data centers, chips, and computer memory. Prices for credit default swaps, remember those popular tools used to bet against corporate debt, tied to Oracle, SpaceX, Alphabet, Amazon, Meta, Broadcom, and Nvidia have risen to record highs in recent days, according to LSEG data. The Sharp moves echo a sell-off in debt issued by so-called
Starting point is 00:10:27 hyperscalers, which are piling hundreds of billions of dollars into developing vast data centers and sophisticated AI models. It comes as investors have grown increasingly worried about the deluge of debt sold by these companies. Credit markets don't deal well with uncertainty and the sheer unpredictability of the pace and cost of AI financing is triggering a serious crisis of confidence right now, said John Oller Ward, chief director of the credit management sona asset management firm. In a sign of waning investor interest in AI debt, Meta's latest borrowing costs for its $12 billion Texas Data Center has risen significantly to levels closer to junk-rated bonds. The debt, quote, priced in line with where B-minus deals are currently trading,
Starting point is 00:11:13 a quite remarkable situation, but today that's the world we are living in, Al-Award said. The moves have been most acute at Oracle, whose five-year CDS was quoted at 215 basis points on Monday up from 144 basis points at the start of the year, meaning that investors now need to pay $215,000 annually to ensure $10 million of debt against default. The Database Group last month said it would invest $70 billion in the coming year to finance its data center buildout, prompting the S&P Global Rating to downgrade its credit rating to triple B minus, just one notch above junk status, citing an uncertain path to profitability amid massive AI investments.
Starting point is 00:11:59 The big question is, will this level of capital spending grow in the number of, in perpetuity and when is that inflection point where we will see positive cash flow again, said David Brown, global co-head of investment grade at Nureberger Bergman. We won't have the answer anytime soon, which explains the weakness and performance. It's potentially going to be a problem because there's still so much financing waiting to be done, Brown said, end quote. This episode is brought to you by Accenture. When your advertising operations fall out of sync, Everything else follows. Spotify and Accenture are working together to reinvent the rhythm of ad sales, using automation, analytics, and smarter workflows to simplify campaign delivery and access better data across the business.
Starting point is 00:12:50 The result? Less time spent on operations, more time connecting brands with the moments and fandoms that matter most. Learn more at Accenture.com slash Spotify. When critical company knowledge isn't documented, there's a major ripple effect. work becomes inconsistent. Tools don't get adopted and knowledge walks out the door when someone leaves. Thankfully, our sponsor, Scribe, was built to fix that. Their workflow AI platform is trusted by nearly half of the Fortune 500 to capture workflows in real time. Here's how it works. You turn on the Scribe browser extension or desktop app, do a process as you normally would, and Scribe will build a guide as you go. It automatically redacts sensitive information and even suggests improvements to
Starting point is 00:13:33 your existing workflows. To see what Scribe could look like for your your org, head to scribe.how slash ride home and mention ride home for your first month of Scribe capture free on select plans. That's S-C-R-I-B-E dot how slash ride home. It's a common misconception that nicotine can be a pre-workout, but it's actually a vaso-constrictor detrimental to workout. So, to help get the pump you're looking for, ultra pouches can give you instant focus and an energy boost with zero nicotine. It uses clinically. proven new tropics and adaptogens to deliver immediate focus and energy that lasts one to two hours. Just pop a pouch in five minutes before your workout. You can feel it sharp in your mind
Starting point is 00:14:19 and lift energy by the time you start. New customers can use code TBRH to get 15% off at Take Ultra.com. That's take ultra.com for 15% off with code TBRH. Real quick, Apple has in fact launched that Apple upgrade program I told you about earlier. Here are the details from Mac rumors, quote, Apple upgrade has lower base prices than the iPhone upgrade program with iPhones available starting at 1799 per month and the Apple Watch available starting at 1199 per month. Max can be leased starting at 2499 per month and iPads start at 1199 per month. AppleCare Plus is optional and not included in the lease price with customers also able to opt for AppleCare 1. There are 12-month and 24-month leasing options for the
Starting point is 00:15:11 iPhone and Apple Watch, along with 24-month and 36-month leasing options for the Mac and iPads. Lease costs varied based on devices and is lower with a device trade-in that's applied on a monthly basis. Apple has some simple pricing minus taxes and trading credits. I'm actually not going to list these all. I'm going to quote again, when leasing an iPhone, customers are required to choose a plan from AT&T, Verizon, or T-Mobile, and prepaid plans are not eligible. iPhones need to be leased with a carrier plan, but the iPhones are unlocked so customers can switch carriers if desired. Mv&Os, like MintMobil or Visible, are not supported. At the end of the leasing period, customers can choose to return the device and exit the program,
Starting point is 00:15:56 pay off the remaining amount owed on the device with a one-time payment and keep it, or return it, and upgrade to a new device with a new lease. The payoff amount is the difference between what was paid during the leasing period and the retail price of the device, minus any remaining trading credits. Klarna is not charging a fee for leasing this program as a partner with Apple. So an iPhone, that's $1,99 can be leased and then purchased for $1,09 with no extra cost beyond taxes. As with trade-ins, Apple will send a pre-labeled and prepaid shipping box for device returns or upgrades. If you don't opt to pay the purchase fee at the end of the leasing program, you will not own the device,
Starting point is 00:16:38 and you must return it, end quote. Finally today, this is a public service announcement from me to you in this Inoregum period where we know AI can hack stuff, but people are still trying to give only the ammunition to the white hats for as long as possible. So, any time any of your devices tells you of a security update, you should run it. Run, don't walk. to run it and install it. For example, got an Apple device. There's a new update that has, let me check, 155 CVEs in it, quoting 9 to 5 Mac. Apple fixed more than a dozen kernel
Starting point is 00:17:24 vulnerabilities. The potential impacts included corrupting or writing to kernel memory, disclosing kernel memory, bypassing network filters, and causing unexpected system termination. WebKit received a sizable collection of fixes as well. One additional Wi-Fi vulnerability could allow a nearby hacker to corrupt process memory. In addition to many fixes shared with iPhone and iPad, the Mac update addresses vulnerabilities that could let apps gain root access, escape their sandboxes, bypass gatekeeper checks, or privacy preferences, and access protected data. Apple has also released security updates for older MacOS versions today, but those are not included in the counts here. Apple has also detailed fixes for watchOS 26.6 TVOS and VisionOS.
Starting point is 00:18:09 Apple's latest updates address 194 unique CVEs after removing overlap between platforms. Apple previously shared that it accelerated the release of a number of security fixes for last month's iOS 26.5.2 release due to AI-powered hacking tools,
Starting point is 00:18:26 end quote. Nothing more for you today. Talk to you tomorrow. Hey, y'all. It's Kelly Clarkson with Wayfair. Ever order furniture online and wonder what if, like what if it doesn't hold up? That sofa was four days old. You should have ordered from Wayfair. With Wayfair, there's no what if.
Starting point is 00:18:49 Just style you love and quality you can trust. Visit Wayfair.ca. Wayfair, every style, every home.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.