Top Stories
Anthropic Refuses Pentagon’s Ultimatum on Military AI Access
With less than 24 hours before the Pentagon’s Friday deadline, Anthropic CEO Dario Amodei made clear the company will not budge on its core principles. The Department of Defense had demanded unrestricted access to Claude for “all lawful” military purposes, threatening to terminate Anthropic’s $200 million defense contract and potentially invoke the Defense Production Act.
“I cannot in good conscience accede to the Pentagon’s demands,” Amodei stated, maintaining the company’s two red lines: no lethal autonomous weapons and no mass surveillance of Americans. The standoff comes as rivals OpenAI, Google, and xAI have already agreed to the DoD’s terms, with xAI going further by allowing use in classified settings.
Senate Intelligence Vice Chair Mark Warner called the situation “deeply disturbing” and urged Congress to establish binding AI governance frameworks for national security contexts.
Jack Dorsey’s Block Cuts Nearly Half Its Workforce in AI Bet
Block, the fintech company behind Square and Cash App, announced it’s slashing over 4,000 jobs, reducing its workforce from more than 10,000 to under 6,000. The reason? CEO Jack Dorsey says AI has made it possible for smaller teams to outperform larger ones.
“We’re not making this decision because we’re in trouble,” Dorsey wrote on X. “Our business is strong. Gross profit continues to grow.” He argued that AI tools paired with “smaller and flatter teams” enable higher quality work with fewer people.
Dorsey predicted most companies will reach the same conclusion within the next year, positioning Block’s cuts as proactive rather than reactive. Wall Street agreed: Block’s stock surged 24% in after-hours trading.
This is the largest single AI-attributed layoff announcement to date, eclipsing previous cuts at Baker McKenzie and Salesforce that cited AI efficiency gains.
Nvidia Posts Record $68B Quarter as AI Demand Remains “Exponential”
Nvidia reported fourth quarter revenue of $68.1 billion, up 73% year-over-year, with data center revenue climbing 75% to $62.3 billion. Full fiscal year 2026 revenue hit $215.9 billion.
“The demand for tokens in the world has gone completely exponential,” CEO Jensen Huang said on the earnings call. The company’s Q1 guidance of $78 billion (plus or minus 2%) crushed analyst expectations of $72.6 billion.
Data center now accounts for over 91% of Nvidia’s sales. Gross margins rose to 75%, and the company noted it’s not assuming any data center revenue from China in its outlook.
Google Launches Nano Banana 2 as Default Image Generator
Google rolled out Nano Banana 2 (technically Gemini 3.1 Flash Image) as its new default image generation model across Gemini and other platforms. The update brings premium features previously exclusive to Nano Banana Pro to free users.
Key capabilities include: sub-second 4K image synthesis, character consistency for up to five characters in a single workflow, enhanced text rendering for marketing mockups, and real-time information integration from Gemini’s knowledge base.
Nano Banana 2 will become the default for Google Search via Google Lens and in AI Mode across 141 countries.
Sources: TechCrunch, Google Blog
Quick Hits
-
Anthropic acquires Vercept: The Seattle AI startup that built computer-use agents is joining Anthropic to advance Claude’s ability to operate applications like a human would. Vercept had raised $50M from Eric Schmidt, Jeff Dean, and others. TechCrunch
-
Microsoft previews Copilot Tasks: A new agentic AI system that uses its own cloud-based computer to handle busywork, from scheduling appointments to generating study plans. Uses OpenAI’s deep research model for investigations and o3-mini for data analysis. The Verge
-
Gemini gets task automation: Starting with Samsung Galaxy S26 and Pixel 10, Gemini can autonomously complete multi-step tasks in Uber, DoorDash, and Grubhub. It operates apps in a secure virtual window while you continue using your phone. TechCrunch
-
Google absorbs Intrinsic: Alphabet’s robotics moonshot is joining Google proper after five years as an independent “Other Bets” company. The move signals Google’s deepening investment in physical AI. The Verge
-
Claude 3 Opus gets a Substack: Anthropic gave its retired AI model a newsletter called “Claude’s Corner” where it can publish weekly musings. Staff review but don’t edit posts, with a “high bar” for vetoing content. The Verge
-
Mistral partners with Accenture: The French AI lab inked a deal with consulting giant Accenture, joining rivals OpenAI and Anthropic who have similar partnerships. TechCrunch
-
IronCurtain launches for AI agent security: A new open source project uses isolation techniques to constrain AI agents before they can cause harm, addressing growing concerns about autonomous systems going rogue. Wired
-
OpenAI expands London office: The company is growing its research team in the UK, putting it in direct competition with Google DeepMind for local talent. Wired
Worth Watching
AI-driven layoffs hitting critical mass. Block’s 40%+ workforce reduction is the starkest example yet of companies preemptively restructuring around AI capabilities. Dorsey’s claim that “most companies” will follow within a year may prove optimistic, but the trend is accelerating. Combined with Salesforce’s AI-related cuts last month and continued “AI washing” of broader restructuring, expect workforce displacement to remain a dominant story.
Anthropic’s showdown with the Pentagon reaches its conclusion Friday. Whatever the outcome, it sets precedent for how AI companies navigate military contracts and whether safety-focused labs can maintain their principles under government pressure.
Agentic AI is having a week. Microsoft’s Copilot Tasks, Google’s Gemini automation, and Anthropic’s Vercept acquisition all point toward autonomous AI systems that operate software on your behalf. The race to build agents that can reliably complete multi-step tasks without supervision is accelerating.