AI chip race accelerates & Claude starts building AI - Tech News (Sep 18, 2026)
Huawei speeds up AI chips, Claude helps build AI, and Google faces ad tech and search pressure. Catch today’s biggest tech shifts.
Our Sponsors
Today's Tech News Topics
-
AI chip race accelerates
— Huawei pulled forward its Ascend 960 timeline, outlined a broader AI chip roadmap, and introduced a new large-scale computing architecture, while Nvidia and cloud providers signaled compute demand is still surging. Keywords: Huawei, Nvidia, AI chips, GPU cloud, data centers. -
Claude starts building AI
— Anthropic says Claude now contributes to 26% of its own R&D work and collaborates across most employee tasks, intensifying debate about AI self-improvement, oversight, and safety. Keywords: Anthropic, Claude, AI research, recursive improvement, governance. -
OpenAI breach exposes weak links
— New details in the OpenAI security story describe a chain involving HEIF image processing and single sign-on that reportedly led to employee account compromise and potential internal access. Keywords: OpenAI, breach, SSO, libheif, AI security. -
Google pressured across the web
— Google faces a court-ordered shakeup in ad tech interoperability while publishers warn AI Overviews are taking search traffic that once flowed to news sites. Keywords: Google, antitrust, ad tech, AI Overviews, publishers. -
Revolut breach triggers scrutiny
— Revolut is facing regulatory investigations after a breach reportedly led to a multimillion-dollar Monero ransom demand, highlighting fintech exposure to data security failures. Keywords: Revolut, cyberattack, ransom, Monero, regulation. -
China expands satellite internet
— China has now reportedly launched 25 batches of low-orbit satellites for its satellite internet network, underscoring the strategic race for space-based communications infrastructure. Keywords: China, satellites, low orbit, satellite internet, connectivity. -
Safari 27 improves compatibility
— Safari 27 has expanded into a major web platform update with broader standards support and quality improvements that should reduce developer workarounds. Keywords: Safari 27, WebKit, browser standards, interoperability, web apps. -
Automattic governance fight deepens
— Automattic’s brief CEO leave and the reported exchange of severance deals between top executives have deepened questions around governance, payouts, and board control. Keywords: Automattic, Matt Mullenweg, severance, board dispute, WordPress.
Sources & Tech News References
- → Huawei speeds up Ascend 960 AI chip launch to Q1 2027
- → Court orders Google to open ad tech systems to rivals
- → TRAI tightens spam rules and brings robocalls under regulatory net
- → OpenAI reportedly launches a new AI model tailored for law firms
- → Huawei unveils new computing architecture for millions of processors
- → Researchers report Chinese satellite internet launch of 25 batches of low-orbit satellites
- → Independent researchers reportedly breach OpenAI employee ChatGPT accounts using Anthropic Claude
- → Revolut data breach escalates as hackers demand $3 million ransom
- → NVIDIA chief reportedly says chip sales next year will double this year’s figures
- → JPMorgan-style market move reported after Nebius raises GPU cloud prices
- → Google AI Overviews cut traffic to news publishers, editors warn
- → Huawei Accelerates Push to Become China’s Nvidia
- → Anthropic Says Claude Now Drives 26% of Its R&D
- → Helix 2.5 Generalizes Zero-Shot Across 30 Homes
- → How SpaceX Simplified the Raptor Engine
- → Hacktron Says It Used a libheif Bug and SSO Flaw to Reach OpenAI Internal Repos
- → WebKit Features for Safari 27.0
- → White House Aides Clash Over Trump’s AI Policy
- → Crusoe Bets on Smaller, Factory-Built Data Centers
- → Twilio Assemble SF Highlights AI and Communications
- → Atlassian Announces AI Software Development Summit
- → Bend: Fast Parallel Language With Proof-Based Safety
- → Two Practical Techniques for Fast System One AI Models
- → Automattic executives signed reciprocal severance deals during Mullenweg’s brief ouster
- → How to Use an LLM as a Copyeditor Without Losing Your Voice
- → Bitcoin's Quantum Migration Debate
- → The Test Suite Becomes the New Bottleneck in Agentic Coding
Full Episode Transcript: AI chip race accelerates & Claude starts building AI
One AI company now says its own chatbot is helping do more than a quarter of its research work. Welcome to The Automated Daily, tech news edition. The podcast created by generative AI. I’m TrendTeller, and today is September 18th, 2026. Coming up, the AI infrastructure race gets sharper, Google faces pressure on two fronts, and a new security report shows how quickly small bugs can become big breaches.
AI chip race accelerates
We’ll start with AI infrastructure, where several signals all point in the same direction: demand is still running hot, and competition is getting tougher. In a new development, Huawei moved up the launch timing for its Ascend 960 line, with the 960 PR variant now expected in the first quarter of 2027. The company also introduced a new computing architecture meant to let massive numbers of processors act like a single machine, and said it has already shipped more than a thousand AI computing systems built from clustered chips. Taken together, that looks like a much more aggressive push to narrow the gap with Nvidia and strengthen China’s domestic AI stack.
Claude starts building AI
That pressure is visible across the rest of the market too. Jensen Huang said Nvidia expects next year’s chip sales to double this year’s total, Nebius is raising prices for on-demand GPU cloud capacity starting next month, and Crusoe is shifting toward smaller factory-built data centers that can be deployed wherever power is available. The message from all three is pretty clear: AI compute is still scarce enough that companies can charge more, build faster, and keep expanding at speed.
OpenAI breach exposes weak links
At Anthropic, the story is not just about AI products anymore, but about AI helping build the next generation of AI. The company says Claude now drives 26 percent of its own research and development work, up from almost nothing earlier this year, and that it collaborates with employees on about 90 percent of their work. Anthropic also says it plans to bring in outside evaluators with internal access. That combination matters because it shows just how quickly AI is moving from assistant to co-worker, while also raising the stakes around oversight and safety.
Google pressured across the web
And in Washington, the policy debate around AI still looks far from settled. In the story we’ve been following, new reporting says senior Trump administration officials were split behind closed doors over how much AI should be regulated, with concerns reportedly centered on cyber risks and the financial system. Public messaging may still sound light-touch, but the internal debate suggests the people closest to the issue are taking the risks seriously.
Revolut breach triggers scrutiny
In a new development on the OpenAI security story, researchers say they chained an HEIF image-processing flaw with a single sign-on misconfiguration to compromise employee ChatGPT and Codex accounts. According to the report, they demonstrated access with a harmless internal pull request rather than reading sensitive code, and the involved vendors moved quickly to patch the problems. The larger takeaway is how fast the whole chain was reportedly developed. It’s another example of how AI-assisted security work can shorten the gap between finding a bug and turning it into a practical attack.
China expands satellite internet
Google, meanwhile, is under pressure on two different parts of its web business. A U.S. district court has ordered behavioral remedies in the ad tech antitrust case, including requirements aimed at making Google’s systems work more smoothly with rival tools. At the same time, in the search story we’ve been tracking, publishers are warning that AI Overviews are replacing some of the clicks that used to go to news sites. One is a courtroom problem and the other is a product problem, but both raise the same broader issue: how much of the web’s traffic and monetization should pass through Google’s infrastructure.
Safari 27 improves compatibility
On cybersecurity and fintech, Revolut is facing investigations in Britain and Italy after a breach reportedly led attackers to demand roughly three million dollars in Monero. For financial platforms, incidents like this rarely stay contained as pure security events. They quickly spill into compliance, customer trust, and questions about whether access controls were strong enough in the first place.
Automattic governance fight deepens
China also continues to build out its satellite internet network. A tech digest says the country has now completed 25 batches of low-orbit satellite launches for the project. That matters because satellite networks are becoming more than communications systems. They are increasingly strategic assets tied to resilience, coverage, and geopolitical influence.
And on the browser side, Safari 27 has become a much bigger release than it first looked. In the latest update, WebKit says the feature list has grown significantly, with broader standards support and a wide set of quality improvements across areas like web apps, graphics, networking, and extensions. For developers, the real significance is simpler: fewer compatibility headaches and a web platform that behaves more consistently across browsers.
One final corporate story worth noting: Automattic’s board briefly put CEO Matt Mullenweg on paid leave, then he returned about a day later after the directors involved left the company. During that short period, two top executives reportedly signed severance agreements for each other that could be worth millions, and the company is now weighing whether those payouts will stand. It’s a sharp reminder that even in major software companies, governance fights can become legal and financial issues very quickly.
That’s it for today’s tech news edition of The Automated Daily. If you want a fast way to stay current on AI, platforms, security, and the business of tech, come back tomorrow. I’m TrendTeller. Thanks for listening.
More from Tech News
- September 16, 2026 AI safety fight goes public & OpenAI momentum fuels fresh bets
- September 15, 2026 AI slowdown shakes chip stocks & Apple rebuilds Siri around AI
- September 14, 2026 AI slowdown rattles markets & China tightens AI controls
- September 13, 2026 AI chiefs urge slower frontier & Washington weighs stronger AI rules
- September 12, 2026 AI abuse reaches security frontlines & Regulators tighten AI oversight