DAILY INTELLIGENCE BRIEF

The Vault — AI Edition

Friday, May 22, 2026
15 stories
17,247 words
76 min read

Business

5
01

Andrej Karpathy Joins Anthropic to Lead Claude Pre-Training Research Team

In a move that sent a jolt through the AI industry, Andrej Karpathy — OpenAI co-founder, former Tesla AI director — announced he is joining Anthropic.

6 minRead →
02

Japan Megabanks Set to Access Anthropic Claude Mythos as Finance Minister Confirms Deal

Japan's Finance Minister confirmed Friday that MUFG, SMBC, and Mizuho will receive access to Anthropic's Claude Mythos within two weeks.

6 minRead →
03

Anthropic Opens Milan Office as EMEA Revenue Surges Ninefold Year Over Year

Anthropic is planting its flag in Italy this month, opening a Milan office that adds a sixth European city to its international footprint.

4 minRead →
04

Cerebras Soars 68 Percent in Nasdaq Debut Pushing AI Chipmaker to $95 Billion Market Cap

Wall Street agreed on Cerebras. Shares of the AI chipmaker soared 68% on their first day of trading, closing at $331.07.

4 minRead →
05

AI Search Startups Surge as Exa Labs Raises $250 Million at $2.2 Billion Valuation

The race to rewire how the world finds information just got a fresh infusion of capital — and the numbers are staggering.

5 minRead →

Llms Genai

4
06

OpenAI Launches ChatGPT Personal Finance With Plaid Integration Across 12,000 Banks

On May 15, OpenAI launched a personal finance feature inside ChatGPT that lets Pro subscribers connect their bank accounts.

4 minRead →
07

Kore.ai Ships Artemis Platform to Deploy Production AI Agents in Days Instead of Months

Kore.ai launched its Artemis edition of the Agent Platform on May 21, promising enterprises a path from AI experimentation to production.

6 minRead →
08

Camunda Unveils ProcessOS Agentic Operating System for Enterprise AI Orchestration

In front of 1,200 enterprise leaders gathered in Amsterdam, Camunda CEO Jakob Freund unveiled ProcessOS — a new intelligence layer.

7 minRead →
09

Cloudflare Builds High Performance Multi-GPU LLM Infrastructure With Custom Infire Engine

Cloudflare built Infire, a Rust-based inference engine that runs frontier LLMs across multiple GPUs more efficiently than Python alternatives.

5 minRead →

Research

3
10

Chinese AI Models Overtake US Rivals Capturing 61 Percent of Global API Token Traffic

In the week ending February 24, 2026, something quietly historic happened: Chinese AI models overtook US rivals in global API token traffic.

5 minRead →
11

Hadrian Releases Open Source OpenHack Tool for AI-Powered Vulnerability Discovery in Code

When Hadrian's researchers pointed a commodity LLM at open-source applications used by Dutch government agencies, they found hundreds of vulnerabilities.

5 minRead →
12

MOSS Framework Enables AI Agents to Self-Evolve Through Autonomous Source Code Rewriting

An AI agent that can diagnose its own recurring failures, rewrite the source code responsible, and test the patches — all without human help.

5 minRead →

Policy

3
13

Bipartisan Senate Group Introduces FARM AI Act to Expand Artificial Intelligence for American Farmers

A six-senator bipartisan coalition introduced legislation on May 21 that would embed AI research and workforce training into USDA programs.

5 minRead →
14

Meta Tracked Employee Keystrokes and Screenshots to Build AI Training Data Before Mass Layoffs

Before Meta cut roughly 8,000 employees on May 20, it had already been harvesting something else from them: their work patterns and keystrokes.

4 minRead →
15

Japan Forms 36-Entity Public-Private Task Force to Counter AI Cybersecurity Threats

Japan assembled a 36-member cross-sector working group to identify and neutralize systemic risks from AI vulnerability discovery tools.

5 minRead →