2026-09-26 · ~8 min read Past Edition View today's briefing →

10 picked from 60 candidates · ordered by significance

Today's Insight

The harder Washington leans on Anthropic, the bigger the bets Anthropic keeps making.

Anthropic showed up in four different kinds of news on September 25 alone. A DC appeals court ruled 2-1 that the Pentagon's designation of Claude as a supply-chain risk can stand. The same day, the White House asked OpenAI and Anthropic to hold back new models from the UK's until US agencies finish their own review, and Anthropic had already withheld its newest model from the UK.

Continue reading (3) Show less

Yet instead of pulling back, the company kept scaling up. That same day, it signed a new seven-year, $11.6 billion cloud deal with Akamai that could grow to $20 billion if certain conditions are met. It's also tightening its grip on governance: co-founders including Dario Amodei, who each own only about 2% of the company, asked shareholders to approve special stock giving them a combined 50.1% of the vote on most matters before the IPO.

The center of gravity in the consumer AI assistant race is shifting too. Meta's personal agent Muse reportedly outpaced ChatGPT's early adoption numbers. Microsoft is pitching its redesigned Copilot as a 'super app as influential as Office,' but its actual target is enterprises, not consumers, and it quietly dropped the 'Copilot+ PC' hardware brand from its new Surface laptops, likely a response to Recall's security backlash and fading AI PC marketing momentum.

The same day also showed government oversight giving way to industry self-regulation. Google, OpenAI, and Anthropic are working toward launching a voluntary body called SAFA this year to set safety standards for frontier models without direct government involvement. With the Trump administration's regulatory skepticism and congressional gridlock stalling government-led rules, companies are stepping in to fill that gap themselves.

Signal to watch Whether SAFA actually launches this year and brings open-source AI makers into the fold is the next thing to watch.

Appeals Court Lets the Pentagon Designate Anthropic a Supply-Chain Risk

Summary

A federal appeals court in Washington, DC ruled on September 25 that the Pentagon's designation of Anthropic's Claude as a supply-chain risk can stand, in a 2-1 decision. The majority found the Defense Department could treat Anthropic's own restrictions on what Claude will do, refusing to support autonomous weapons or domestic surveillance, as grounds for a national-security risk finding. Defense Secretary Pete Hegseth had framed that refusal as the threat justifying the block.

Why it matters

Why It Matters

This lets the Pentagon keep blocking Claude from government systems indefinitely, with Anthropic's remaining options, a larger appeals panel or the Supreme Court, likely taking years to resolve. In the meantime, rivals like Grok, Gemini, and GPT models are positioned to fill the government-contract gap Claude leaves behind.

Read the original

Also covered by Ars Technica AI

The Pentagon wants $30 million to build an AI-powered lie detector

Summary

The Pentagon's budget request calls for $30.3 million over five years to build an AI-powered lie detector called Polygraph+, run by the Defense Counterintelligence and Security Agency. The program pairs machine-learning scoring algorithms with contactless sensing that reads heart rate, head movement, facial skin temperature, and pore activity through a camera, with no equipment attached to the body. It surfaced around the same time roughly 50 Joint Chiefs officers were reportedly polygraphed after a leak about weapons stockpiles.

Why it matters

Why It Matters

Experts are skeptical: legal scholar Kyri Kotsoglou calls it a case of oversimplifying a complex problem, and Sophie van der Zee argues polygraphs mainly work as a deterrent rather than an accurate readout. The US National Academy of Sciences already rated polygraph evidence as "at best modest" back in 2003, and adding AI scoring on top doesn't erase that underlying limitation.

Read the original

Also covered by The Decoder

Anthropic to pay Akamai $11.6 billion over seven years in cloud deal

Summary

Anthropic has committed $11.6 billion over seven years to Akamai's cloud infrastructure, more than six times the $1.8 billion deal reported back in May, with the total able to grow to $20 billion if certain conditions are met. In exchange, Akamai gave Anthropic a , convertible into 7.7 million shares at $111.33 each for an initial stake of about 2%, that grows by roughly a percentage point for every additional $3 billion Anthropic spends, up to a 5% cap.

Why it matters

Why It Matters

The warrant structure mirrors the one AMD used in its OpenAI deal, showing how cloud providers are increasingly trading future stock upside instead of straight cash for landing a major AI customer. Akamai expects roughly $1.7 billion in annual revenue from this deal by the end of 2028, a shift that could pull a company built on content delivery networks toward becoming an AI infrastructure provider.

Read the original
See it drawn
A cloud company gave up equity, not just cash, to land one AI customer.

Also covered by The Decoder

Meta's Muse just stole the AI spotlight from OpenAI and Anthropic

Summary

In a week when Anthropic launched Claude Opus 5.5 and OpenAI answered with a GPT-6 update just 90 minutes later, it was Meta's personal AI agent Muse that ended up dominating the conversation. Muse reportedly outpaced ChatGPT's early mobile adoption numbers, and Meta plans to extend it into smart glasses and a Tamagotchi-style wearable.

Why it matters

Why It Matters

It's a reminder that winning the model-performance news cycle and becoming the app people actually open every day are two different contests. While OpenAI and Anthropic compete on benchmarks and release timing, Meta is getting ahead on consumer adoption by leaning on wearable hardware as the entry point.

Read the original

Also covered by TechCrunch AIThe Verge AITechCrunch AIThe Decoder

Microsoft stops insisting you need a "Copilot+ PC"

Summary

Microsoft's newly announced Surface Pro and Surface Laptop, both launching October 13, quietly drop the "Copilot+ PC" branding the company has pushed since 2024, even though both devices still meet the program's technical requirements. Surface VP Brett Ostrum told Windows Central the machines "do meet all the requirements" but "are not called Copilot+ PCs."

Why it matters

Why It Matters

The quiet retreat is being read alongside the security backlash Recall, the feature that first defined Copilot+ PCs, drew from privacy advocates, plus a broader cooling of "AI PC" as a marketing hook. Compare that to Nadella touting Copilot+ PCs as 15% of premium laptop sales in early 2025, and the AI label is losing ground in hardware marketing just two years after it was introduced.

Read the original

Anthropic's founders seek voting control ahead of IPO

Summary

Anthropic's seven co-founders, including CEO Dario Amodei, each own only about 2% of the company, but they've asked shareholders to approve a new class of stock giving them a combined 50.1% of the vote on most corporate matters. The shares carry no economic value, only voting power, and stay in effect as long as at least three founders keep their stakes. The move comes as the company prepares for an IPO expected to reflect a new valuation.

Why it matters

Why It Matters

Dario Amodei pledged earlier this year to give away 80% of his wealth, warning that AI-driven wealth concentration could destabilize society, and this request can be read as backing that stance with corporate control: no matter how many outside shareholders join after the IPO, the founders intend to keep steering the company's direction, safety decisions included.

Read the original

Also covered by TechCrunch AI

Microsoft thinks its new Copilot ‘super app’ will be as influential as Office

Summary

Microsoft officially unveiled its redesigned Copilot "super app" on September 25, bundling chat, code generation, and an always-on agent into a single interface with three tabs: Home, Code, and Autopilot. The agent, previously known as Scout, has been renamed Autopilot and keeps running in the cloud even when a user is offline, monitoring Teams and Outlook to keep handling tasks. Home and Code roll out over the coming weeks, while Autopilot starts as a private preview in late September.

Why it matters

Why It Matters

Rather than competing head-on with Meta's Muse or Google's Gemini for personal chatbot users, Microsoft is betting on its Office ecosystem and enterprise customer base. Shifting Autopilot, Code, and Cowork to usage-based billing instead of flat fees also marks a step back from the flat-rate AI subsidies Microsoft has offered until now.

Read the original

White House Pressures OpenAI, Anthropic to Withhold New Models from UK Reviewers

백악관, 오픈AI·앤트로픽에 "신모델, 영국 사전 검증 거부하라" 압박

Summary

Politico reported on September 25 that the White House, at the request of the National Cyber Director, has asked OpenAI and Anthropic to hold back new models from the UK's until US agencies finish their own review. Anthropic has already withheld its newest model from the UK AISI, saying it's currently available only to select US institutions, while the UK AISI's director confirmed pre-release testing cooperation continues on other frontier models.

Why it matters

Why It Matters

This marks the first real gap in the pre-release testing arrangement the UK and frontier AI companies agreed to at the 2024 AI Safety Summit. With the US Commerce Department's own evaluation body, CAISI, already bottlenecked by staff shortages and no confirmed director, prioritizing US-first review could ironically end up delaying international rollout of the newest models.

Read the original

Google, OpenAI, and Anthropic Push to Launch Self-Regulatory Body 'SAFA' This Year

구글·오픈AI·앤트로픽, 자율 규제기구 'SAFA' 연내 출범 추진

Summary

The Information reported on September 24 that Google, OpenAI, and Anthropic are working to launch a voluntary industry body called the Standards Authority for Frontier AI (SAFA), aiming to set safety benchmarks for frontier models without direct government oversight, with a target launch of late 2026 or early 2027. SAFA would handle third-party safety evaluations before model releases and set incident-reporting procedures, with names like former White House AI policy adviser Sriram Krishnan floated for CEO and former Secretary of State Condoleezza Rice for chair.

Why it matters

Why It Matters

With government-led regulation stalled by the Trump administration's skepticism and congressional gridlock ahead of the November midterms, companies are moving to fill that vacuum themselves. But with open-model makers like Meta left out of the initial evaluation scope, critics warn SAFA could double as a barrier that uses safety standards to box out competitors.

Read the original

Can Apple Home’s AI camera features outsmart Amazon’s and Google’s? I put them to the test

Summary

Apple Intelligence for Home, launched this month alongside iOS 27, finally brings AI-generated text descriptions to HomeKit Secure Video, a feature Ring, Nest, Blink, Wyze, and Arlo have already offered. The surprise is that Apple charges up to three times as much as competitors for it. The reviewer tested a Ring Pro 4K, a Google Nest Doorbell, and a HomeKit-compatible Aqara G400 side by side to compare how each AI system describes what its camera sees.

Why it matters

Why It Matters

The gap between a generic "motion detected" alert and "small dog chasing a chicken" can decide how fast someone actually responds. As text alerts that arrive before the video clip loads become standard, the competition among security cameras is shifting from video quality toward how accurately and quickly a camera can tell you what it just saw.

What to do now

If you use HomeKit Secure Video, check your settings to see whether Apple Intelligence notifications are turned on, they may not be included automatically in your existing plan.

Read the original

Terms in this briefing

Past Briefings

Monthly Reports

Weekly Recaps