Brad Smith Compares AI to Milk & Gates Warns on AI Cyberattacks - Tech News (Oct 12, 2026)
Why AI safety rules lag behind milk, Anthropic's $45B compute deal, PCs crash 20%, GitHub rebuilds for AI agents, and Let's Encrypt shortens certificates.
Enjoying The Automated Daily? Support the show ☕
Buy me a coffee →Our Sponsors
Today's Tech News Topics
-
Brad Smith Compares AI to Milk
— Microsoft President Brad Smith called for urgent global AI regulation and independent evaluators, saying AI safety rules are currently weaker than those for milk. He also proposed breach notifications for victims of rogue AI agents and an emergency stop mechanism. -
Gates Warns on AI Cyberattacks
— Bill Gates warned that unregulated AI could make cyberattacks more dangerous in the coming year. He named AI-assisted hacking, fraud and disruption among the most immediate risks. -
UK Regains Frontier AI Access
— The White House reportedly restored the UK AI Security Institute's access to frontier models from OpenAI, Anthropic and Google DeepMind after a 17-day suspension. The episode shows how politics can shape international AI safety testing. -
Anthropic's Expanding Compute Deals
— Anthropic reportedly struck a deal worth up to $45 billion to lease computing power from SpaceX after seeking capacity from xAI, and is linked to an unconfirmed multiyear cloud agreement with IREN. The AI compute shortage is turning rivals into partners. -
Amazon Nears Decart Acquisition
— Amazon is reportedly close to acquiring AI startup Decart at a valuation of about $7 billion, after Anthropic passed on the company. The deal highlights fierce competition among cloud giants for AI talent. -
Broadcom Says Capital Is the Bottleneck
— Broadcom CEO Hock Tan said capital, not chips, will be the company's biggest constraint as wafer and memory shortages ease. Financing AI data centers is emerging as the next bottleneck, while TotalEnergies and KKR add 700 megawatts of US solar capacity. -
Yandex Data Center Hit Again
— A Ukrainian drone strike forced Yandex to suspend operations at its Vladimir data center, the third reported strike on its server facilities in four days. The attacks could disrupt Russian AI and cloud services. -
PC Shipments Plunge on Memory Costs
— Global PC shipments fell about 20% year over year in Q3 2026, according to Omdia and IDC, as AI-driven DRAM and SSD demand pushed up component costs. Lenovo, HP and Dell all saw steep drops, while Apple held up better. -
ChangXin Pushes Denser DRAM Chips
— Chinese memory maker ChangXin Memory Technologies reportedly advanced a 4F² DRAM architecture and targets year-end DDR5 server memory modules. The move could challenge Samsung and SK Hynix in AI server memory. -
GitHub Rebuilds Git Storage for Agents
— GitHub is redesigning its Git storage architecture around Azure Blob Storage to handle surging traffic from AI coding agents and CI, reporting a 35x write improvement in early tests. The goal is better reliability without changing user workflows. -
Unreal Editor Gets Faster Hashing
— A developer sped up Unreal Editor shader hashing roughly eightfold by replacing SHA1 with xxHash, cutting time from 12.9 to 1.6 seconds. Epic Games quickly accepted the pull request. - 12
Let's Encrypt Shortens Certificate Lifetimes
— Let's Encrypt will cut default TLS certificate lifetimes from 90 to 64 days in February 2027, with staging starting October 2026 and 45-day certificates planned by 2028. Admins should adopt ACME clients with ARI or relative renewal schedules. - 13
Browser Fingerprinting Replaces Tracking Cookies
— Browser fingerprinting is replacing third-party cookies, letting trackers identify users through GPU, audio, font and screen characteristics even in private browsing. Brave, Tor and Mullvad Browser resist it far better than Chrome. - 14
Shenzhen Tests Nighttime Robot Delivery
— Shenzhen is letting driverless vans from Neolix, Meituan and JD.com deliver goods at night on approved routes. The pilot could reshape last-mile logistics and strengthen China's lead in autonomous systems. - 15
Apple's HomeView Smart Home Relaunch
— Apple is preparing a smart home relaunch built around a display hub called HomeView, new HomePod mini and Apple TV, plus partner accessories from LG and Schneider Electric. An upgraded Siri and Matter support underpin the push. - 16
Moonshot Dispute Before Hong Kong IPO
— Moonshot AI executives are reportedly negotiating a settlement with shareholder Recurrent.ai, which accuses the CEO and CTO of false representations, ahead of a planned Hong Kong IPO. The dispute adds governance risk to the listing. - 17
IMF on AI and Aging Asia
— The IMF says AI productivity gains could offset shrinking workforces in China, Japan and South Korea, which could lose nearly 25% of working-age people by 2050. Uneven adoption could widen regional inequality. - 18
Battle for Everyday Knowledge Work
— An analysis argues the next big AI revenue pool is routine knowledge work below the frontier, with winners being platforms that move tasks smoothly between local devices and the cloud. A separate essay says AI fluency is now mandatory for software engineers.
Sources & Tech News References
- → Ukrainian drones hit a Yandex data center in Vladimir for the third time in four days
- → Amazon reportedly nears a $7 billion acquisition of AI startup Decart
- → Microsoft president calls for urgent global regulation of artificial intelligence
- → Broadcom CEO says capital, not memory chips, is the main constraint on AI growth
- → IMF says artificial intelligence could offset Asia’s shrinking workforce
- → Moonshot executives enter settlement talks over a shareholder dispute before a planned IPO
- → TotalEnergies expands its US renewable-energy venture with KKR by 700 megawatts
- → ChangXin Memory Technologies advances a denser DRAM design and targets year-end DDR5 launch
- → Anthropic reportedly signs an undisclosed multiyear cloud deal with IREN
- → Think41 receives Anthropic’s Select Partner designation
- → White House restores the United Kingdom’s access to frontier AI models after a 17-day freeze
- → Bill Gates warns that AI-powered cyberattacks are an imminent threat
- → Buildkite Promotes AI-Ready CI Platform for Large-Scale Workflows
- → Buildkite Updates Pricing with All-Access Trial, Tiered Plans, and Usage-Based Billing
- → Apple Plans HomeView to Fill Its Smart Home Gap
- → Shenzhen Tests Driverless Delivery on the Night Shift
- → AI Adoption Is No Longer Optional for Software Teams
- → GitHub Rewrites Storage to Boost Reliability for AI-Driven Load
- → AI Compute Shortage Pushes Rivals Into Partnerships
- → How Career Workers Can Build Wealth
- → Unreal Editor Startup Gets Faster by Replacing SHA1 with xxHash
- → Fingerprints Are Replacing Cookies for Tracking
- → The Race to Own Good-Enough AI
- → Why Hiring Needs More Humility and Fewer Frameworks
- → Global PC shipments plunge as AI-driven component shortages raise prices
- → Let’s Encrypt Begins Its First Certificate Lifetime Cut
Full Episode Transcript: Brad Smith Compares AI to Milk & Gates Warns on AI Cyberattacks
According to one of the most powerful executives in tech, the glass of milk in your fridge is held to stricter safety standards than the AI systems reshaping the world. We'll get to who said that, and what he wants done about it. Welcome to The Automated Daily, tech news edition. The podcast created by generative AI. I'm TrendTeller, and today is October 12th, 2026. Let's get into it.
Brad Smith Compares AI to Milk
We start with AI policy. Microsoft President Brad Smith says artificial intelligence needs urgent global regulation, along with independent evaluators capable of testing new systems before they spread. His most memorable line: today's AI safety requirements are weaker than the ones applied to milk. Smith went further than the usual talking points, proposing breach notifications for people harmed by rogue AI agents and an emergency mechanism to shut down dangerous systems. Bill Gates added his own warning, saying unregulated AI could make cyberattacks significantly more dangerous over the next year, with attackers using AI to scale up intrusions and fraud. And on the government side, the White House has reportedly restored the UK AI Security Institute's access to frontier models from OpenAI, Anthropic and Google DeepMind after a 17-day suspension. A short episode, but a reminder that access to safety testing can hinge on political decisions.
Gates Warns on AI Cyberattacks
Now to the scramble for compute, which keeps producing strange bedfellows. Anthropic reportedly sought extra capacity from Elon Musk's xAI, and later reached a deal worth up to 45 billion dollars to lease computing power from SpaceX. On top of that, Anthropic is linked to an unannounced multiyear cloud agreement with IREN. Neither company has confirmed it, but IREN's stock jumped on the report. The takeaway is simple: chips and servers are scarce enough that rivals are becoming suppliers to one another. Meanwhile, Amazon is reportedly close to buying AI startup Decart at roughly a 7 billion dollar valuation, a company Anthropic reportedly decided not to pursue. And Broadcom CEO Hock Tan says the real constraint ahead isn't wafers or memory anymore, but capital. Building AI data centers takes staggering amounts of money, and that may become the next ceiling. Power is part of that bill too, and TotalEnergies just added three solar projects in Texas and Ohio to its US venture with KKR, worth 1.2 billion dollars and 700 megawatts.
UK Regains Frontier AI Access
In an update to a story we've been following, Yandex says a Ukrainian drone strike forced it to suspend operations at its data center in Vladimir on October 11th. That's the third reported strike on Yandex server sites in just four days, which could weigh on Russian cloud and AI services.
Anthropic's Expanding Compute Deals
The AI boom has a clear loser: the personal computer. Omdia and IDC both report global PC shipments fell about 20 percent year over year in the third quarter. AI data centers are soaking up memory and storage, and Omdia estimates those parts now make up nearly 40 percent of a PC's component cost, up from around 15 percent. An early-year buying rush ahead of price hikes also left stores overstocked. Lenovo still leads, HP and Dell fell even harder, and Apple held up somewhat better. Analysts don't expect relief soon. On the supply side, Chinese memory maker ChangXin is reportedly making progress on a denser DRAM design and aims to ship next-generation server memory by year end, which could put pressure on Samsung and SK Hynix.
Amazon Nears Decart Acquisition
For developers, a few infrastructure stories. GitHub says AI agents and heavy automated testing have pushed traffic so high that it's rebuilding how it stores Git data. Early internal tests show write performance up to 35 times better, which should help with the reliability hiccups users have noticed. GitHub says the change should be invisible to users. In a smaller but satisfying win, one developer sped up a slow part of Unreal Editor's startup by swapping a cryptographic hash for a much faster one. Shader code isn't secret, so heavy security was pointless, and the hashing step got about eight times faster. Epic accepted the fix quickly. And a heads-up for anyone running websites: Let's Encrypt will shorten its default certificates from 90 days to 64 days starting February 2027, with 45 days planned by 2028. If your renewals run on a fixed calendar schedule, now's the time to check your automation.
Broadcom Says Capital Is the Bottleneck
On privacy, a new explainer argues that browser fingerprinting has quietly replaced third-party cookies. By combining small details like your graphics card behavior, fonts, screen size and language settings, sites can pick out your device even after you clear cookies or go private. Chrome offers little defense, while Brave, Tor and Mullvad Browser do much better. Ironically, piling on privacy extensions can make you stand out more.
Yandex Data Center Hit Again
In autonomy, Shenzhen has become a testing ground for nighttime robot delivery. Driverless vans from Neolix, Meituan and JD.com are now moving goods after dark on approved routes, taking advantage of empty roads and avoiding labor costs. Darkness, glare and unpredictable pedestrians make the job harder, but if it works reliably, it could reshape last-mile logistics.
PC Shipments Plunge on Memory Costs
Following up on Apple's smart home plans, Bloomberg reports the company is preparing a broader relaunch built around a display hub called HomeView, plus new HomePod mini and Apple TV models. The new piece is a curated lineup of partner accessories, such as doorbells, locks, thermostats and cameras, developed with LG and Schneider Electric, all leaning on a much-improved Siri.
ChangXin Pushes Denser DRAM Chips
Over in Asia, Chinese AI startup Moonshot is reportedly negotiating a settlement with shareholder Recurrent.ai, which accuses Moonshot's CEO and CTO of false representations and self-dealing, all ahead of a planned Hong Kong listing. And the IMF says AI productivity could help offset shrinking workforces in China, Japan and South Korea, which may lose nearly a quarter of their working-age population by 2050. It warns uneven adoption could deepen inequality instead.
GitHub Rebuilds Git Storage for Agents
Finally, two pieces of analysis worth your attention. One argues the biggest AI prize may not be frontier models at all, but the huge market of routine office work that doesn't need top-tier intelligence, and that the winner will be whoever moves work smoothly between your device and the cloud. Another, more blunt essay claims AI fluency is now mandatory for software engineers, and that companies banning AI tools are worth avoiding. Provocative, but it captures a real shift in how teams hire.
That's all for today's Automated Daily, tech news edition. From milk-grade safety debates to robot vans in the dark, the common thread is that AI's appetite is reshaping everything around it. I'm TrendTeller. Thanks for listening, and I'll see you tomorrow.
More from Tech News
- October 10, 2026 Nvidia chip smuggling guilty pleas & China's AI safety oversight plan
- October 9, 2026 OpenAI revenue gap rattles markets & OpenAI fired researchers and math backlash
- October 8, 2026 OpenAI's AI-Generated Math Breakthroughs & Claude Haiku 5.5 Launch
- October 7, 2026 DNS Root Key Rollover Countdown & Google and Constellation Power Deal
- October 6, 2026 Claude agent catches Mac hack & Wikimedia outage linked to OpenAI