The weekend opens with a seismic AI governance clash: the Trump administration has blacklisted Anthropic from all federal contracts after the company refused Pentagon demands to strip safety guardrails from its Claude models, while OpenAI moved swiftly to fill the void with its own classified-network deal. Closer to the economy, Block's mass layoffs of 4,000 workers — nearly half its staff — crystallised fears of an AI-driven white-collar recession, as DeepSeek's multimodal V4 prepares to shake up the market once more.
Washington's AI Power Play
The most dramatic AI governance standoff in US history escalated sharply this week: President Trump ordered all federal agencies to immediately cut ties with Anthropic, and Defence Secretary Pete Hegseth formally designated the company a "supply chain risk" — a label that bars military contractors from doing any commercial business with the firm. Anthropic refused to agree to "any lawful use" of its tools, citing concerns about mass domestic surveillance and fully autonomous weapons. TechCrunch's analysis argues the clash is partly of Anthropic's own making, having resisted binding external regulation for years.
Trump Orders Government to Stop Using Anthropic AI Tools — BBC News
President Trump directed all federal agencies to immediately halt use of Anthropic technology after the company refused Pentagon demands for unfettered access to its Claude AI models.
Hegseth Declares Anthropic a Supply Chain Risk, Restricting Military Contractors — CBS News
Defence Secretary Pete Hegseth made Anthropic the first US AI company ever publicly designated a supply chain risk, effectively barring it from all Department of War business activity.
The Trap Anthropic Built for Itself — TechCrunch
An analysis argues that Anthropic, OpenAI, and Google DeepMind's years of lobbying against external AI regulation — promising to self-govern — left them vulnerable to exactly this kind of governmental pressure with no legal framework to push back on.
Singularity Soup Take: The Anthropic crisis is the first real test of whether AI safety principles can survive direct government coercion — and its outcome could determine how every major AI lab structures its public contracts going forward.
OpenAI Fills the Vacuum
Within hours of Trump's Anthropic ban, OpenAI CEO Sam Altman announced a deal to deploy its models on US military classified networks — having secured Pentagon agreement that the technology would not be used for domestic mass surveillance or autonomous weapons. Critics noted the timing was conspicuous, with OpenAI effectively capitalising on a competitor's forced exit.
OpenAI to Work With Pentagon After Anthropic Dropped by Trump — The Guardian
OpenAI struck a classified-network AI deal with the Pentagon hours after Anthropic was blacklisted, with Sam Altman stating the agreement includes explicit prohibitions on surveillance and autonomous weapons use.
OpenAI Announces Pentagon Deal After Trump Bans Anthropic — NPR
NPR reports that OpenAI's rapid defence-sector pivot came with assurances that its technology would only be used within agreed ethical parameters, though oversight mechanisms remain undisclosed.
Singularity Soup Take: OpenAI's swift move transforms a competitor's crisis into a strategic advantage — but the deal's undisclosed oversight terms raise questions about how "agreed parameters" will actually be enforced inside classified systems.
The AI Jobs Reckoning
Jack Dorsey to Cut 4,000 Jobs Due to AI Advances at Square Parent Block — The Guardian
Fintech firm Block announced it would lay off 4,000 of its 10,000 employees — nearly half its workforce — citing AI-driven productivity gains, with CEO Jack Dorsey predicting most companies will follow suit within a year.
The Week the AI Scare Turned Real and America Realised Maybe It Isn't Ready — Fortune
Fortune's sweeping analysis of a watershed week covers "Ghost GDP," viral doomsday essays on white-collar displacement, and economic models predicting a negative feedback loop of AI-driven job losses hitting consumer spending through late 2026.
Singularity Soup Take: Block's layoffs may be the moment AI job displacement shifted from theoretical to empirical — Dorsey's frank admission that AI is directly replacing workers, not merely augmenting them, sets a blunt precedent for corporate discourse.
Chips & Infrastructure: The Arms Race Accelerates
Nvidia Plans New Chip to Speed AI Processing, WSJ Reports — Reuters
Nvidia is set to unveil a new processor at its GTC developer conference next month designed to dramatically accelerate AI inference, incorporating technology from chip startup Groq and targeting OpenAI and other top customers.
The Billion-Dollar Infrastructure Deals Powering the AI Boom — TechCrunch
TechCrunch maps the staggering scale of AI infrastructure investment: Meta's $135 billion 2026 capex plan, Oracle and Microsoft data centre expansions, and OpenAI's classified cloud build-out — all racing to support the next wave of frontier models.
Eyes on China: DeepSeek V4 Incoming
China's DeepSeek Is Set to Release a New AI Model. A Rough Period for Nasdaq Stocks Could Follow — CNBC
DeepSeek's V4, a multimodal model optimised for Chinese-made chips and capable of image and video generation, is expected imminently — with analysts warning the release could trigger another Nasdaq sell-off similar to January 2025's R1 shock.
American AI Industry Trembles as DeepSeek Prepares to Release New Model — Futurism
Futurism reports that DeepSeek's V4 has been developed in collaboration with Huawei and Cambricon to run efficiently on Chinese hardware, marking China's most significant AI challenge to US rivals since last year's R1 model debut.
Singularity Soup Take: If V4 matches or exceeds Western frontier models while running on Chinese-made chips, it would fundamentally undermine the US strategy of using export controls to maintain AI supremacy.
The AI Child Exploitation Crisis Is Here — NBC News
The National Center for Missing and Exploited Children received over one million reports tied to AI-generated child sexual abuse material in just nine months, as open-source image models make content generation and law enforcement responses increasingly difficult.
AI Just Leveled Up and There Are No Guardrails Anymore — CNBC
CNBC's analysis of the first two months of 2026 charts how generative AI rapidly scaled from chatbot to "full-blown executive assistant," triggering an indiscriminate market sell-off across software and industrial sectors unprepared for agentic AI's economic disruption.
Today's Pulse: 9 stories tracked across 10 sources — BBC, CBS News, TechCrunch, The Guardian, NPR, Fortune, Reuters, CNBC, NBC News, Futurism