Why CERN is rebuilding the LHC

The Frontier

Science & tech breakthroughs in AI models, hardware, and computing that define what's possible

The disconnection of the Large Hadron Collider has begun. CERN has started disconnecting sections of the Large Hadron Collider to replace beam-focusing magnets for the High-Luminosity LHC upgrade. The replacement magnets will be about 40% stronger, supporting a higher collision rate when the upgraded accelerator begins operating.

CERN is betting the High-Luminosity LHC’s payoff comes from 40% stronger focusing magnets, not higher beam energy. The test is whether the replacements deliver a higher collision rate when the upgraded accelerator begins operating.

Scitechdaily

Google tests buying from Walmart-owned Flipkart through Gemini and AI Mode. In India, Google is testing a purchase flow for selected Flipkart products within Gemini and Google AI Mode. The test adds a Buy button that routes users to Flipkart checkout, initially for a limited group of users and products including smartphones, electronics and accessories.

Google is positioning Gemini and AI Mode as a shopping layer in India, handing checkout to Flipkart rather than building its own. Watch for the test to expand beyond selected electronics and smartphones by year-end.

TechCrunch

Anthropic’s blacklisting by the Pentagon was legal, federal judges rule. A federal appeals court in Washington ruled that the Pentagon’s blacklisting of Anthropic was legal. The court said the Defense Department had ample support for its conclusion that Anthropic’s products could pose a national-security risk.

Anthropic's loss in the D.C. Circuit locks its Pentagon ban for the current 5-year procurement cycle. Courts have upheld only three similar tech blacklists since 2018: Huawei, Kaspersky, DJI. Net effect: OpenAI and Palantir now face fewer bidders for FY27 AI contracts.

New York Times

Model Developments

Latest AI model releases, capabilities, and technical breakthroughs shaping the competitive landscape

Oxford lets OpenAI train its AI models on the Bodleian Library. The University of Oxford allowed OpenAI to digitise historical texts from the Bodleian Library for use in its training set, according to internal documents reviewed by The Guardian. University staff raised concerns about reputational risk as model developers seek new sources of digitised academic material.

OpenAI’s Bodleian access shows Oxford is willing to supply historical texts for AI training despite staff reputational concerns. Watch for whether Oxford authorises another such access arrangement by year-end.

The Guardian

Anthropic turns Claude into an AI marketplace with 2,000+ plugins and connectors. Anthropic announced a Claude Marketplace offering more than 2,000 plugins, connectors, agents and related tools. The marketplace brings third-party integrations around Claude into one place, expanding the product from a standalone model interface into a distribution layer for enterprise workflows and external services.

Anthropic has made Claude an enterprise distribution layer, but a 2,000-plus catalog alone does not prove marketplace traction. By year-end, the test is whether Anthropic reports a higher catalog total and identifies whether plugins, connectors or agents are driving growth.

Bleepingcomputer

Capital & Control

Tracking the flow of capital and influence that shapes the tech landscape

Apple hit with $5.7 billion in damages over haptic patents. Apple was ordered to pay $5.7 billion in damages after its Taptic Engine was found to infringe two patents held by Taction. The verdict creates a substantial patent liability tied to a core hardware interface used across Apple devices.

Apple’s $5.7B verdict—about 6% of 2025 net income—makes Taction’s haptic patents a meaningful licensing lever, not an existential threat to Apple. The test is whether Android OEMs seek pre-emptive licenses before holiday 2027 cycles.

The Verge

TikTok agrees to pay at least $100M in Alabama settlement. TikTok agreed to pay Alabama at least $100 million to settle allegations that it misled users about platform safety and was designed to addict children. Alabama said payments could reach $300 million under specified conditions, while TikTok also agreed to add a two-hour daily limit, parental controls, overnight-use restrictions and limits on cosmetic filters for underage users.

TikTok’s settlement puts a $100M floor under child-safety compliance, with $300M at stake under specified conditions. By 2027, the test is binary: are the two-hour limit, parental controls, overnight restrictions and underage filter curbs in place?

TechCrunch

Infrastructure & Power

The physical backbone of AI: compute, data centers, supply chains, and the geopolitics of energy

Cloudflare fixes Containers cross-tenant flaw exposing customer data. Cloudflare fixed a cross-tenant vulnerability in its Containers and Sandboxes products that could allow Workers Paid customers to recover residual data from other customers’ containers sharing a physical host. The issue involved isolation in multi-tenant cloud infrastructure rather than an application-level exposure.

Cloudflare’s fix addresses a serious multi-tenant isolation breach, not an app defect: Workers Paid customers could recover residual data across tenants. In the next Containers or Sandboxes update, watch for evidence that cross-tenant residual-data recovery is blocked.

Bleepingcomputer

Digital Defense

Threats and defenses shaping the global security landscape

ShinyHunters uses WAF bypass trick in Oracle PeopleSoft attacks. The ShinyHunters extortion group is using a URL-encoding technique to bypass web application firewall rules intended to mitigate Oracle PeopleSoft vulnerability CVE-2026-35273. The bypass has enabled renewed widespread exploitation of vulnerable PeopleSoft servers.

ShinyHunters has turned WAF mitigation into a weak control: vulnerable PeopleSoft servers remain exposed when URL-encoded requests evade rules. The test by end-2026 is whether patching and request normalization curb renewed widespread exploitation.

Bleepingcomputer

The Performance Edge

Peak performance science for sustained execution in high-stakes environments

Atrial fibrillation patients saw 69% lower risk with common blood thinners. The randomized SINGLE-AF trial found that direct oral anticoagulants reduced serious clinical events by 69% among atrial-fibrillation patients at intermediate stroke risk. The reduction was driven mainly by fewer ischemic strokes, with no increase in major bleeding reported.

69% reduction in serious events makes no-anticoagulant management hard to justify for intermediate-risk AF patients; fewer ischemic strokes came without more major bleeding. Watch for ESC guidance on wider use within 2 years.

Science Daily

Market Pulse

Daily market data: indices, sectors, top movers, and volatility

Markets closed for the weekend. US equity markets are closed. The next market print arrives Monday at 6:30am Pacific, when the NYSE opening bell rings.

NYSE’s next market print arrives Monday at 6:30am Pacific, making the opening the only near-term price signal after the weekend closure. Watch for that print Monday; it will set the week’s first observable market level.

Muninn