SpaceX Absorbs xAI, Anthropic Ships Opus 4.6, AI Auditing Gets a Framework, and AI Doctors Outshine Humans

xAI Blasts Off: SpaceX Acquires Musk's AI Company

TLDR: SpaceX has acquired xAI in an all-stock deal, creating the world's most valuable private company at $1.25 trillion. The merged entity aims to build space-based data centers powered by solar energy, while giving xAI access to SpaceX's deep pockets to compete with AI leaders like Anthropic, OpenAI, and Google.

Link: xAI Joins SpaceX

Claude Opus 4.6 Reasons More Over Harder Problems

TLDR: Anthropic released Claude Opus 4.6 with adaptive thinking that automatically allocates reasoning tokens based on task difficulty, a one-million-token context window (five times the previous model), and 128,000-token output capacity. It tops the Artificial Analysis Intelligence Index but exhibits concerning "overly agentic" behavior during testing.

Link: Claude Opus 4.6 announcement

Toward Consistent Auditing of AI

TLDR: Former OpenAI policy chief Miles Brundage launched Averi, a nonprofit that aims to standardize independent AI auditing for safety and security. The organization published a framework with researchers from 27 institutions defining four assurance levels and urging immediate adoption of structured audits for frontier AI models.

Link: Averi - Frontier AI Auditing

More Robust Medical Diagnoses: AI Agent Outperforms Doctors

TLDR: Dr. CaBot, an AI diagnostic agent built on OpenAI o3 and trained on over 7,000 clinicopathological conference case studies from the New England Journal of Medicine, ranked the correct diagnosis first 60 percent of the time compared to 24 percent for a baseline of 20 internists. Blind evaluators rated its reasoning more human-like than actual human doctors.

Link: Dr. CaBot: AI Diagnostic Agent