Today's AI news is dominated by a high-stakes standoff between Anthropic and the Pentagon over Claude's military use limits, alongside fresh accusations that Chinese AI labs secretly distilled Claude's capabilities using tens of thousands of fake accounts. Markets remain rattled by a viral doomsday scenario report as Nvidia prepares to deliver its biggest quarterly earnings yet, and Samsung has unveiled the AI-forward Galaxy S26 series at Unpacked in San Francisco.

Anthropic vs. The Pentagon: Safety Lines in the Sand

Defence Secretary Pete Hegseth met with Anthropic CEO Dario Amodei at the Pentagon on Tuesday, issuing a Friday deadline: agree to unrestricted military use of Claude — including for mass surveillance and lethal autonomous weapons — or face being designated a "supply chain risk," a classification normally reserved for foreign adversaries. Alternatively, the Pentagon threatened to invoke the Defense Production Act to compel compliance. Anthropic has refused, drawing a bright line against its AI being used for autonomous weapons or mass surveillance of Americans.

Anthropic Won't Budge as Pentagon Escalates AI Dispute — TechCrunch
The Pentagon has given Anthropic until Friday to loosen Claude's safety guardrails or face potential penalties, including the Defence Production Act — a significant expansion of the law's use that legal experts call unprecedented.

Inside Anthropic's Existential Negotiations with the Pentagon — The Verge
The dispute centres on three words — "any lawful use" — terms OpenAI and xAI have reportedly already accepted, which would give the US military carte blanche for surveillance and lethal autonomous AI systems with no human in the loop.

Why it matters: This is arguably the most consequential AI governance standoff in the US to date — a company's safety principles versus federal power, with Anthropic's entire defence and enterprise business on the line if labelled a national security threat.

Anthropic Accuses Chinese AI Labs of Secretly Training on Claude

Anthropic has accused three Chinese AI companies — DeepSeek, Moonshot AI, and MiniMax — of setting up more than 24,000 fake accounts and generating over 16 million exchanges with Claude to extract and distil its capabilities, targeting its most advanced agentic reasoning and coding skills. The accusations emerged alongside the Pentagon dispute, reinforcing Anthropic's argument for tighter AI chip export controls to China.

Anthropic Accuses Chinese AI Labs of Mining Claude as US Debates AI Chip Exports — TechCrunch
Anthropic says DeepSeek, Moonshot, and MiniMax used 24,000 fake accounts in a large-scale "distillation attack," and argues that restricting chip exports to China would limit the scale of such illicit model harvesting.

Anthropic Accuses DeepSeek and Other Chinese Firms of Using Claude to Train Their AI — The Verge
DeepSeek alone held over 150,000 exchanges with Claude specifically targeting its reasoning capabilities, and is accused of using those extractions to generate censorship-safe alternatives to politically sensitive questions.

Why it matters: If confirmed, this represents one of the most documented cases of AI model theft at scale — and hands US policymakers a concrete argument for stricter export controls on advanced chips to China.

Viral AI Doomsday Report Sends Markets Sliding

A 7,000-word hypothetical scenario from Citrini Research — titled the "2028 Global Intelligence Crisis" — went viral this week, portraying a near-future in which autonomous AI agents gut white-collar jobs, collapse software valuations, private credit, and mortgages, pushing US unemployment above 10% by mid-2028. Despite being labelled a speculation, not a prediction, the Dow Jones Industrial Average fell 1.7% (822 points) on Monday, with companies named in the report — including Uber, Mastercard, and American Express — each losing 4–6%.

'A Feedback Loop with No Brake': How an AI Doomsday Report Shook US Markets — The Guardian
Citrini Research's speculative 7,000-word scenario — portraying AI agents creating an unstoppable economic collapse from software firms through to mortgages — rattled Wall Street and sent the S&P's software index to its lowest since Trump's Liberation Day tariff shock.

AI-Linked Fears Roil Some Corners of Wall Street After Years of Hype and Gains — NBC News
Some investors now fear AI is too good at certain tasks and is already causing permanent disruption to software-heavy industries, reflecting a sharp shift in market sentiment from AI optimism to AI anxiety.

Why it matters: Markets are increasingly pricing in not just AI opportunity but AI disruption risk — a psychological shift that could reshape investment flows in tech and finance regardless of whether any specific scenario plays out.

Nvidia's Biggest Earnings Test Yet

Nvidia is set to report Q4 fiscal 2026 earnings after market close today (Wednesday, 25 February), with Wall Street projecting $65.9 billion in revenue — up 66.7% year over year — and earnings per share of $1.52. The report lands at a fraught moment: Nvidia's stock is barely up 2% in 2026 while most of the tech index has fallen, and competition from custom silicon at Meta, Amazon, and Google is intensifying. Analysts say the results serve as the central bellwether for the entire AI infrastructure trade.

Nvidia Earnings Report Collides with Wall Street Skepticism Over AI Spending — CNBC
While hyperscalers continue spending heavily on AI infrastructure, investors are scrutinising whether Nvidia can sustain its dominant position as in-house chip programmes from Amazon, Google, and Meta mature and multiply.

Nvidia's Q4 Earnings Report Marks a Pivotal Moment for the AI Trade — Business Insider
With $60 billion of Q4 revenue expected from data centers alone, tonight's results will either confirm or complicate the central thesis underpinning three years of AI market exuberance.

India AI Impact Summit — The New Delhi Declaration

The India AI Impact Summit wrapped last weekend with 88 countries, including the US, China, Russia, the UK, and the EU, signing the New Delhi Declaration on AI Impact — a non-binding agreement aimed at making AI work for the Global South, underrepresented languages, and inclusive economic development. The summit drew $200 billion in investment commitments, including a $110 billion pledge from Reliance's Mukesh Ambani. However, tension remained: the White House delegation made clear the US "totally rejects global governance of AI."

India AI Summit 2026 Marks New Delhi's Strong AI Beginning, Not Its Arrival — Yet — Indian Express
Despite the historic New Delhi Declaration signed by 88 nations and tech CEOs including Sam Altman and Sundar Pichai in attendance, analysts say India's ambitions to become a global AI powerhouse will require years of execution, not just declarations.

India's AI Impact Summit Closes with the New Delhi Declaration and a $200 Billion Boost — Fortune
The summit's headline outcome included commitments to strengthen testing and evaluation of AI systems across underrepresented languages, with special focus on making frontier AI models more reliable across the Global South.

Why it matters: The New Delhi Declaration is the broadest international AI agreement since the Bletchley Park declaration, and the US–China participation alongside the White House's rejection of binding global AI governance reveals deep fractures beneath the surface consensus.

Samsung Galaxy S26 Unveiled at Unpacked

Samsung held its Galaxy Unpacked event in San Francisco today, announcing the Galaxy S26, S26 Plus, and S26 Ultra — its new flagship smartphone lineup built around Galaxy AI. Highlights include AI-powered camera enhancements, a Zero-Peeking Privacy screen feature, text-based AI image editing, Perplexity integration, and an upgraded Bixby voice assistant. The launch positions Samsung directly against Apple's AI-integrated devices in the premium smartphone market.

Samsung Galaxy Unpacked 2026: The Galaxy S26 Series, AI and Other Products — Engadget
Samsung's Unpacked event centres on the Galaxy S26 series, with Galaxy AI features deeply embedded into the camera system and core phone UI, aiming to make on-device intelligence feel native rather than bolted on.

Samsung Galaxy Unpacked 2026: What to Expect, How to Watch, Live Updates — CNET
Samsung's Galaxy S26 Ultra is expected to overhaul its camera sensor stack to close the gap with iPhone and Google Pixel, while new Galaxy Buds 4 Pro and updated Galaxy Watch round out the AI-focused product lineup.

Meta's $100 Billion Deal Buys It Freedom from Nvidia Dependence — Marketplace
Meta and AMD have announced a multiyear deal worth up to $100 billion for AI chips, with AMD issuing Meta a warrant for up to 10% of its stock — a novel chips-for-equity structure designed to accelerate Meta's move away from Nvidia's GPU monopoly.

AI Offers Range of Possible Futures for Unemployment, Stock Market — Axios
From cautious optimism to worst-case economic disruption, analysts examining near-term AI impact on jobs and markets say the sheer divergence of possible outcomes is itself the defining feature of the current moment.

Generative AI Analyzes Medical Data Faster Than Human Research Teams — ScienceDaily
A Wayne State University study found that generative AI can process massive medical datasets significantly faster than traditional computer science research teams, and in some cases produce stronger analytical outputs.

Today's Pulse: 9 stories tracked across 13 sources — TechCrunch, The Verge, The Guardian, NBC News, CNBC, Business Insider, Indian Express, Fortune, Engadget, CNET, Marketplace, Axios, ScienceDaily