← Back to archive

Monday, May 18, 2026

Anthropic beats OpenAI (and Foxconn gets wrecked)

Anthropic is quietly eating OpenAI's lunch in enterprise adoption while OpenAI pivots to putting Codex in your pocket with new mobile apps. Meanwhile, Foxconn got absolutely hammered by ransomware—8TB of data stolen including files from Apple, Nvidia, Google, and Intel (yikes). Oh, and academics just figured out how to generate high-quality images in four diffusion steps instead of dozens. Would you bet on the scrappy underdog or the household name?

Top Stories

1
NEXUS: An Agentic Framework for Time Series Forecasting

NEXUS is a multi-agent LLM framework for time series forecasting that separates macro and micro temporal analysis before synthesis. Testing on 2025 data beyond model training cutoffs, it matches or beats specialized Time Series Foundation Models while providing explainable reasoning.

llmagentstime-seriesforecasting
2
OpenAI brings Codex to mobile

OpenAI is bringing its Codex desktop AI coding tool to ChatGPT mobile apps on iOS and Android, allowing remote control and management from phones. The move is part of OpenAI's effort to compete with Anthropic's Claude Code and advance its desktop superapp strategy.

openaicodexmobileai-agents
3
Foxconn Confirms Nitrogen Ransomware Attack on North American Plants — Apple, Nvidia, Google, Intel Data Allegedly Stolen in 8TB Breach

The Register

Foxconn confirmed a major ransomware attack on North American facilities with 8TB of data allegedly stolen, potentially exposing information from Apple, Nvidia, Google, and Intel. The incident underscores critical supply chain cybersecurity vulnerabilities in the electronics manufacturing sector.

cybersecurityransomwaresupply-chainfoxconn
4
Anthropic beats OpenAI on business adoption

EconLab

Anthropic overtook OpenAI in business adoption for the first time (34.4% vs 32.3%), but faces critical challenges including rising costs, compute constraints, and performance issues that may enable cheaper alternatives to capture market share.

anthropicopenaienterprise-ai-adoptionllm
5
Trajectory Models for Few-Step Diffusion

arXiv

Normalizing Trajectory Models (NTM) achieve competitive text-to-image generation in just four steps while maintaining exact likelihood training, addressing a key limitation of existing few-step diffusion methods that sacrifice the likelihood framework through distillation or adversarial approaches.

diffusiongenerative-aicomputer-visiontext-to-image

Keep Reading

Enjoyed this issue?

Get daily AI intel delivered to your inbox. No fluff, just the stories that matter.