AI News Roundup: Claude Code Source Leak, Oracle Cuts 30K Jobs, GPT-5.4 Goes Agentic

Anthropic Accidentally Leaks Claude Code’s Entire Source Code

In one of the most embarrassing incidents in AI lab history, Anthropic accidentally published the complete source code for Claude Code — its flagship AI coding agent — through a routine npm package update. Version 2.1.88, released on March 31, contained a source map file that exposed nearly 2,000 TypeScript files and over 512,000 lines of code.

The leaked codebase revealed dozens of unshipped feature flags, including a “persistent assistant” background mode, remote control capabilities from a phone or browser, and session review tools. Competitors now have a detailed look at Anthropic’s unreleased product roadmap. Anthropic called it “a release packaging issue caused by human error, not a security breach,” and confirmed no customer data or credentials were exposed.

The irony is hard to miss: the company that brands itself as the safety-first AI lab fumbled basic release packaging. Security researchers and open-source enthusiasts had already forked and mirrored the code within hours before Anthropic could pull it. The incident raises operational security questions at a company now valued at over $60 billion.

Oracle Slashes 30,000 Jobs to Fund $156 Billion AI Bet

Oracle began the largest layoff in its 47-year history on March 31, cutting an estimated 20,000 to 30,000 employees — roughly 18% of its 162,000-person global workforce. Termination emails arrived at 6 a.m. local time across the U.S., India, Canada, Mexico, and Uruguay with no prior warning and immediate system access revocation.

The reason: Oracle is cannibalizing headcount to fund a $156 billion AI data center buildout. TD Cowen estimates the cuts will free $8–$10 billion in annual cash flow. Despite a 95% jump in Q3 net income to $6.13 billion, Oracle took on $58 billion in new debt in just two months and saw free cash flow turn deeply negative at minus $10 billion last quarter.

CEO Clay Magouyrk framed the layoffs as a permanent structural shift, arguing that AI tools now enable smaller teams to produce more software. It’s a stark reminder that the AI boom is simultaneously creating and destroying jobs at unprecedented scale.

Q1 2026 Venture Funding Shatters All Records at $297 Billion

Global venture capital investment hit an all-time high of $297 billion across 6,000 startups in Q1 2026, up approximately 150% both quarter-over-quarter and year-over-year. AI startups captured a staggering 81% of all funding.

Four of the five largest venture rounds ever recorded closed this quarter: OpenAI ($120B), Anthropic ($30B), xAI ($20B), and Waymo ($16B) — collectively raising $186 billion, or 64% of global Q1 investment. OpenAI’s round valued the company at $852 billion, and the company now generates over $2 billion per month in revenue as it prepares for a potential IPO as soon as late 2026.

GPT-5.4 Brings Native Computer Use and 1M Token Context

OpenAI’s GPT-5.4, launched in early March, continues to reshape the agentic AI landscape. It’s the first general-purpose model with native computer-use capabilities, enabling agents to operate desktop applications and execute multi-step workflows autonomously. On the OSWorld-V benchmark for real desktop productivity tasks, GPT-5.4 scored 75% — slightly above the human baseline of 72.4%.

The model supports a 1-million-token context window (roughly 750,000 words or 1,500 pages), enabling developers to process entire codebases or document sets in a single request. Pricing scales with usage: prompts over 272K input tokens are charged at 2x input and 1.5x output rates. Alongside GPT-5.4, reports indicate OpenAI is already finalizing its next model, codenamed “Spud,” expected to launch as GPT-5.5 or GPT-6.

NVIDIA Unveils Rubin Platform and DLSS 5 at GTC 2026

NVIDIA used GTC 2026 to announce the Rubin platform — six new chips designed to power the next generation of AI supercomputers, including the Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch. Rubin is in full production, with partner products arriving in the second half of 2026.

On the gaming side, NVIDIA revealed DLSS 5, calling it the most significant graphics breakthrough since real-time ray tracing. DLSS 5 uses 3D-guided neural rendering and generative AI to produce photorealistic 4K video in real time. However, it’s RTX 5000-exclusive and won’t ship until fall — and NVIDIA is reportedly skipping gaming GPU releases in 2026 entirely to prioritize AI chip production.

EU Moves to Simplify AI Rules as Public Skepticism Grows

The EU Council agreed its position on streamlining AI regulations as part of the “Omnibus VII” legislative package, aiming to simplify the bloc’s digital legislative framework and harmonize AI rules. The move comes as governments try to balance innovation incentives with growing public concern.

A recent Quinnipiac poll found that more than half of U.S. adults believe AI is likely to cause harm, citing job displacement and privacy violations as top concerns. Meanwhile, Singapore is taking a different tack with its National AI Impact Programme, committing to train 100,000 workers in AI skills by 2029 and launching sector-specific AI fluency programs for accountants and lawyers in the first half of 2026.