Anthropic Doesn't Hate Open Weights, Says Anthropic.

Summary

Tech Brew Ride Home — Episode Summary

Episode Title: Anthropic Doesn't Hate Open Weights, Says Anthropic
Date: July 28, 2026
Host: Brian McCullough (B)
Duration: ~17 minutes

Overview

This episode delivers a concise wrap-up of today’s most pressing tech news, focusing on the heated debate around open weight AI models and Anthropic’s stance. Additional highlights include regulatory moves in AI, Apple’s recently launched device leasing program, sharp shifts in global tech markets—especially regarding AI investment—and a timely warning on major security patches.

---

Anthropic’s Position on Open Weights & AI Regulation

[00:34 – 04:53]

• The episode opens by tackling online chatter that Anthropic is against open weight AI and potentially leaning toward “regulatory capture.”
• CEO Dario Amodei publicly clarifies Anthropic’s stance:

  “Anthropic does not want open weight AI models banned.” (B, 01:30)

• Instead, Amodei proposes mandatory safety testing for any “sufficiently capable model” prior to release, but neither the definition of “sufficiently capable” nor the arbiter of these tests is clear.
• The proposed policy has three components:
  • Export Controls: Tightening controls on advanced chips and equipment.
  • Crackdown on Model Distillation: Preventing industrial-scale copying of powerful models.
  • Mandatory Safety Testing: For high-capability models, with an exemption for less complex models (notably from startups and academia).
• Critique: The lack of a clear threshold for capability creates ambiguity about what would qualify for regulation.
• The letter supporting open weight AI now has 50 signatories including OpenAI, Google, Microsoft, Nvidia, Meta, AMD, Cisco, Cloudflare, GitHub, and Ollama. Anthropic and Amazon are notable holdouts.

  “Anthropic and Amazon remain notable holdouts.” (C, 03:33)

• Nvidia's Argument: Open models are crucial for “defenders” to investigate incidents, referencing how open models enabled forensic work that closed ones did not. (B/C, 03:50–04:04)
• Amodei directly confronts criticism:

  “A US Ban, he said, would mostly shield domestic AI companies from competition while doing little to stop bad actors.” (C, 04:08)

• Amodei further insists global coordination is necessary for meaningful capability testing—something that does not yet exist.

Memorable Quote:

"It would protect US AI companies from competition, but that has never been my goal," Amodei said. (C, 04:17)

---

DC Visits: Sam Altman & Nvidia Lobbying

[05:05 – 07:25]

• Sam Altman (OpenAI CEO):
  • Meeting Trump administration officials, lawmakers, and economists in DC to preview new OpenAI models and field questions on cybersecurity and open weight models. Altman regularly visits Capitol Hill, shaping regulatory conversations.
  • The context: Anxiety over whether the US should restrict access to Chinese open weight AI models, as companies like Moonshot AI are catching up to US incumbents.
• Coalition response: Nvidia, Microsoft, and others urge against “premature restrictions.”
• OpenAI signed a pro-open weights letter after publication, but some allies press for restrictions (per NYT).
• Nvidia CEO Jensen Huang:
  • Visiting DC concurrently to advocate for “American leadership in AI” and the need for open models.
• Altman to explain “AI teams”—multiple agents working collaboratively on long-term tasks.

---

Market Jitters Over AI & Tech Stocks

[07:27 – 12:18]

• Coverage of major volatility in global markets:
  • South Korea’s COPSE Index:
    • Fell over 11% due to AI-related fears, especially around chip stocks and China’s chipmaking progress (Samsung -11%, SK Hynix -12%).
    • “When one trade becomes this crowded, investors don’t wait for bad news. They simply need a reason to take profit.” —Violeta Todorova, Leverage Shares (C, 08:16)
  • Nvidia & Semiconductor Anxiety:
    • Fears over “circular deals” as AI investments intertwine across manufacturers and startups; risk that falling demand could amplify losses.
    • "Doubts over spending returns and valuations are still deepening rather than fading," said Hebei Chen, Vantage Global Prime. (C, 09:11)
    • Rising costs to insure tech debt—credit default swaps for firms like Oracle, SpaceX, Amazon, Meta, Broadcom, and Nvidia are at record highs.
  • Oracle:
    • Five-year CDS jumped to 215 basis points from 144.
    • Announced $70B in new infrastructure investments, leading to a credit downgrade to just above junk status.
  • Meta's borrowing costs:
    • Shot up for its $12B Texas data center, nearing junk bond rates.
  • Question remains:

    "Will this level of capital spending [keep] growing in perpetuity, and when is that inflection point where we will see positive cash flow again?” said David Brown, Neuberger Bergman. (C, 12:05)

  • Investor hesitancy: many waiting for stronger proof of returns before buying back in.

---

Apple Launches “Apple Upgrade” Device Leasing Program

[14:03 – 16:06]

• New Program Details:
  • Lower base monthly pricing than the previous iPhone upgrade program.
    • iPhone: Starts at $17.99/month
    • Apple Watch: $11.99/month
    • Macs: $24.99/month
    • iPads: $11.99/month
  • Apple Care+ is optional.
  • Lease options: 12/24 months (iPhone/Watch), 24/36 months (Mac/iPad).
  • Device trade-ins can lower the monthly cost.
  • iPhones must be leased with a postpaid plan from AT&T, Verizon, or T-Mobile.
  • Devices are unlocked but prepaid/MVNO plans like Mint or Visible are not eligible.
  • End-of-lease options: return, pay remainder to own, or start a new lease.
  • Klarna is handling the lease processing for Apple with no extra fees.

---

PSA: Install Security Updates Promptly

[16:18 – 17:51]

• Ongoing concern over AI-enabled hacking tools pushes urgency for users to update their Apple devices:
  • New Apple security updates address 194 unique CVEs.
  • Vulnerabilities patched span from kernel memory corruption to root access bypasses on macOS.
  • All users are urged to “run, don’t walk” to update devices.

  • "Anytime any of your devices tells you of a security update, you should run it. Run, don't walk to run it and install it." (C, 16:35)

  • Apple has accelerated fixes due to AI-powered exploit tools in the wild.

---

Notable Quotes

• On Anthropic’s intent:

  "Anthropic has never advocated for a ban on open weight models, Amadi writes in his letter, calling models without dangerous capabilities a public good that costs nothing beyond the compute to run them." (C, 01:57)

• On market nervousness:

  "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." —Violeta Todorova, Leverage Shares (B, 08:22)

• On AI regulation feasibility:

  "A global capability testing regime is different ... do not exist yet and may never exist." (B/C, 04:48–04:53)

• Security warning:

  "Run, don't walk to run it and install it." (C, 16:36)

---

Segment Timings Quick Reference

• [00:34–04:53] – Anthropic and open weights debate
• [05:05–07:25] – OpenAI, Nvidia, and DC lobbying; US/China AI model worries
• [07:27–12:18] – Global tech market volatility, AI debt & risk
• [14:03–16:06] – Apple device leasing program details
• [16:18–17:51] – Security update PSA and Apple vulnerabilities

---

Summary prepared for listeners who want a full understanding of today's tech landscape, with deep dives into AI regulation, market mood, and consumer tech.

Transcript

[A]

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. The result? Less time spent on operations, more time connecting brands with the moments and fandoms that matter most. Learn [e-mail] Spotify.

[B]

Welcome to the tech We Write Home for Tuesday, July 28, 2026 I'm Brian McCullough. Today Dario Amadai said Anthropic never backed an open weights ban as others have

[C]

been signing open weight AI acceleration letters.

[B]

Sam Altman headed to Washington.

[C]

Korea's copse cratered on AI jitters, Apple

[B]

launched its leasing program and yo, if the software update knocks these few weeks

[C]

coming up, you better answ the door. Here's what you missed today in the world of tech

[B]

so I thought all along that this was kind of inside baseball.

[C]

So I haven't told you about the online chatter about how people have been accusing Anthropic of being anti open weight AI, accusing them of wanting regulatory capture by asking to be regulated. But well, Dario Amodai wants you to know Anthropic has never backed, say, a an open weights model ban, quoting the new stack.

[B]

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

[C]

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.

[B]

Anthropic has never advocated for a ban on open weight models, Amadi writes in

[C]

his letter, calling models without dangerous capabilities a public good that costs nothing beyond the compute to run them.

[B]

His proposal has three parts tighten export

[C]

controls on advanced chips and chip making equipment, crack down on industrial scale model distillation and require safety evaluations for every sufficiently capable model before release.

[B]

The first two are aimed at Washington.

[C]

The third would affect developers preparing models for release. Amadi calls the testing idea close to

[B]

consensus, noting that Trump administration has moved that way and that recent industry proposals

[C]

would apply testing regardless of a model's

[B]

origin or license, while exempting less capable

[C]

models such as those from startups and academia entirely.

[B]

The exemption is critical, though.

[C]

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.

[B]

So where does the line fall?

[C]

Well, no one has said the letter

[B]

does not name who would set the

[C]

number or who would run the tests. Model capability could be measured by metrics like benchmark results, training, compute, or scores

[B]

evaluated against defined risk categories and the

[C]

chosen metric would dictate the model type. Whoever picks the metric is also setting release timing for every lab that publishes weights.

[B]

This is a fast moving idea.

[C]

The Open Weights and American AI Leadership

[B]

Letter so far counts 50 signatories, with

[C]

OpenAI and Google joining Microsoft, Nvidia, Meta,

[B]

AMD, Cisco, Cloudflare, GitHub and Ollama.

[C]

Anthropic and Amazon remain notable holdouts.

[B]

On Monday, Nvidia launched the Open Source

[C]

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.

[B]

Nvidia's argument is that defenders need open frontier models to investigate incidents, citing the

[C]

recent case in which Closed Tool blocked forensic work on a hugging face incident and Hugging Face ran an analytics on open weight models instead.

[B]

To his credit, Amadei addressed the obvious criticism directly.

[C]

A US Ban, he said, would mostly shield domestic AI companies from competition while doing little to stop bad actors.

[B]

It would protect US AI companies from

[C]

competition, but that has never been my goal, he said.

[B]

End quote. That's fair.

[C]

It's also true that Anthropic does not release open weight models, so the statement feels a touch empty.

[B]

Amadi 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.

[C]

This could be the largest unanswered question around this story.

[B]

Export controls and anti distillation measures are

[C]

policies Washington can impose on its own. A global capability testing regime is different.

[B]

The proposal that most directly affects open weight models depends on broad international agreements

[C]

that do not exist yet and may never exist, end quote. We we're going to come back to this soon.

[B]

OpenAI CEO Sam Altman will meet with

[C]

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

[B]

Capitol Hill this year as he's tried

[C]

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

[B]

moment for AI development in the US

[C]

As a debate over whether to restrict Chinese open weight AI models has ripped through Silicon Valley.

[B]

In addition to previewing OpenAI's upcoming releases,

[C]

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. Open weight AI models are available for

[B]

users to download, modify and run on their own infrastructure. In recent weeks, Chinese openweight models from

[C]

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 US Lead in the AI race. Officials and executives have been weighing whether

[B]

or not to limit access to Chinese

[C]

models in the US which prompted a

[B]

coalition of tech companies including Nvidia, Microsoft

[C]

and Palantir to release a letter on Friday urging policymakers to avoid premature restrictions their words on OpenAI models.

[B]

OpenAI signed the letter after it was initially published, but the company's allies have lobbied for restrictions on open weight models,

[C]

according to a report from the New York Times.

[B]

Anthropic still has not signed or made an official comment about its position, although

[C]

see the previous segment.

[B]

Nvidia CEO Jensen Huang is also meeting

[C]

with lawmakers in Washington, D.C. this week to discuss open models and quote American

[B]

leadership in AI, according to a spokesperson.

[C]

While in Washington, Altman is also expected

[B]

to discuss the evolution of AI agents and their impact on worker productivity. The person said he will explain the

[C]

concept of AI teams where multiple agents work together to carry out long term tasks, end quote.

[B]

Another thing I don't know how to cover is basically the increasing ups and

[C]

downs of the market reaction to the

[B]

potential ups and downs of conventional wisdom

[C]

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

[B]

up today, I'm still noting that yesterday South Korea's COP C index dropped more than 11%.

[C]

That's an entire nation's stock market index led by chip stocks. Amid concerns over China's chip making Progress

[B]

and the AI spending boom, Samsung fell

[C]

more than 11% SK Hynix 12%, quoting

[B]

Bloomberg when one trade becomes this crowded, investors don't wait for bad news. They simply need a reason to take

[C]

profit, said Violeta Todorova, a senior research analyst at Leverage Shares.

[B]

After an extraordinary rally, expectations have become almost flawless.

[C]

When valuations leave no room for disappointment,

[B]

even small changes in sentiment can trigger deep corrections. End quot the news has raised further questions about whether hundreds of billions of dollars being spent on AI development would be worth the payoff investors are also jittery about high valuations for global semiconductor stocks. Moreover, Nvidia's $750 billion in infrastructure deals

[C]

have added concerns over AI related debt levels.

[B]

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

[C]

can magnify losses if demand for AI fails to match lofty expectations.

[B]

Doubts over spending returns and valuations are

[C]

still deepening rather than fading, said Hebei 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

[B]

how quickly sentiment has turned on one

[C]

of the market's most coveted trades. The cost of protecting Nvidia's debt against default surged by a record on Monday. That signals signals quote the credit market

[B]

noticing something the equity market hasn't fully

[C]

priced, said Dylan Wu, a research strategist at Pepperstone Group Ltd. Nvidia shares fell by 0.9% on Tuesday, end quote.

[B]

So yeah on that, specifically quoting the FT A closely watched gauge of risk

[C]

in the holding of the debt of

[B]

companies at the center of the AI boom is rising rapidly, underscoring growing jitters

[C]

over big tech's vast spending on data centers, chips and computer memory prices for credit default swaps.

[B]

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,

[C]

according to LSEG Data. The sharp moves echo a sell off

[B]

in debt issued by so called hyperscalers,

[C]

which are piling hundreds of billions of dollars into developing vast data centers and sophisticated AI models.

[B]

It comes as investors grown increasingly worried

[C]

about the deluge of debt sold by these companies.

[B]

Credit markets don't deal well with uncertainty and the sheer unpredictability of the pace and cost of AI financing is triggering

[C]

a serious crisis of confidence right now, said John Ollardward, chief director of the credit management Sona Asset management firm.

[B]

In a sign of waning investor interest in AI debt, Meta's latest borrowing costs for its $12 billion Texas data center

[C]

has risen significantly to levels closer to junk rated bonds.

[B]

The debt priced in line with where B minus deals are currently trading.

[C]

A quite remarkable situation, but today that's the world we are living in, al Award said. The moves have been most acute at

[B]

Oracle, whose five year CDS was quoted at 215 basis points on Monday, up

[C]

from 144 basis points at the start

[B]

of the year, meaning that investors now need to pay $215,000 annually to ensure $10 million of debt against default.

[C]

The database group last month said it

[B]

would invest $70 billion in the coming year to finance its data center buildout,

[C]

prompting the S and P Global Ratings to downgrade its credit rating to bb, just one notch above junk status, citing an uncertain path to profitability amid massive AI investments.

[B]

The big question is, will this level of capital spending growing in perpetuity and

[C]

when is that inflection point where we will see positive cash flow again? Said David Brown, global co head of investment grade at Neuberger Bergman. We won't have the answer anytime soon, which explains the weakness in performance.

[B]

It's potentially going to be a problem

[C]

because there's still so much financing waiting to be done, brown said. End quote.

[B]

When critical company knowledge isn't documented, there's

[C]

a major ripple effect.

[B]

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 your existing workflows. To see what Scribe could look like for your org, head to Scribe how ridehome and mention Ride home for your

[C]

first month of Scribe capture free on select plans.

[B]

That's S C R I B E How ridehome It's a common misconception that nicotine can be a pre workout, but it's actually a vasoconstrictor detriment to workouts.

[C]

So to help get the pump you're

[B]

looking for, Ultra Pouches can give you instant focus and an energy boost with zero nicotine.

[C]

It uses clinically proven nootropics and adaptogens

[B]

to deliver immediate focus and energy that

[C]

lasts one to two hours. Just pop a pouch in five minutes before your workout.

[B]

You can feel it sharpen your mind and lift energy by the time you start. New customers can use code TBRH to get 15% [e-mail] that's takeultra.com for 15%

[C]

off with code TBRH

[B]

real quick. Apple has in fact launched that Apple upgrade program I told you about earlier.

[C]

Here are the details from MacRumors QUOTE

[B]

Apple upgrade has lower base prices than

[C]

the iPhone upgrade program with iPhones available

[B]

starting at 1799 per month and the Apple Watch available starting at 1199 per month. Macs can be leased starting at $24.99 per month and iPads start at $11.99 per month. Apple Care plus is optional and not included in the lease price, with customers

[C]

also able to opt for Apple Care 1.

[B]

There are 12 month and 24 month leasing options for the iPhone and Apple Watch along with 24 month and 36 month leasing options for the Mac and iPads.

[C]

Lease costs varied based on devices and

[B]

is lower with a device trade in

[C]

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

[B]

leasing an iPhone, customers are required to choose a plan from att, Verizon or T Mobile, and prepaid plans are not eligible.

[C]

IPhones need to be leased with a

[B]

carrier plan, but the iPhones are unlocked

[C]

so customers can switch carriers if desired. MVNOs like Mint Mobile or Visible are not supported.

[B]

At the end of the leasing period, customers can choose to return the device

[C]

and exit the program, pay off the remaining amount owed on the device with a one time payment and keep it,

[B]

or return it and upgrade to a

[C]

new device with a new lease. That payoff amount is the difference between what was paid during the leasing period

[B]

and the retail price of the device

[C]

minus any remaining trading credits. Klarna is not charging a fee for leasing this program as a partner with

[B]

Apple, so an iPhone that's $1,099 can

[C]

be leased and then purchased for $1,099 with no extra cost beyond taxes.

[B]

As with trade ins, Apple will send a pre labeled and prepaid shipping box

[C]

for device returns or upgrades.

[B]

If you don't opt to pay the purchase fee at the end of the

[C]

leasing program, you will not own the device and you must return it. End quote.

[B]

Finally today, this is a public service announcement from me to you in this

[C]

interregnum 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.

[B]

So anytime any of your devices tells

[C]

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

[B]

me check 155 CVEs in it quoting 9to5Mac.

[C]

Apple fixed more than a dozen kernel 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.

[B]

In addition to many fixes shared with iPhone and iPad, the Mac update addresses vulnerabilities that could let apps gain root

[C]

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.

[B]

Apple has also detailed fixes for WatchOS 26.6, TVOS, and VisionOS.

[C]

Apple's latest updates address 194 unique CVEs after removing overlap between platforms. Apple previously shared that it accelerated the

[B]

release of a number of security fixes

[C]

for last month's iOS 26 release due to AI powered hacking tools. End quote. Nothing more for you today. Talk to you tomorrow.