The numbers tell a different story than the headlines. Anthropic published a detailed report accusing three Chinese AI labs—DeepSeek, Moonshot AI, and MiniMax—of using roughly 24,000 fraudulent accounts to run 16 million model-distillation exchanges against Claude. Most coverage led with DeepSeek. We dug into the per-lab breakdown: according to Anthropic’s data, MiniMax accounted for roughly 79% of the total exchange volume. Moonshot AI contributed around 21%. DeepSeek’s share was under 1%. The gap between the headline name and the actual volume leader is worth understanding.
While Anthropic deals with the distillation fallout, OpenAI is making a major enterprise push. The company just formalized Frontier Alliances with McKinsey, BCG, Accenture, and Capgemini to deploy its agentic AI platform across Fortune 500 clients. Anthropic currently draws roughly 80% of its revenue from enterprise—exactly the market OpenAI is now targeting with four of the largest consulting firms on the planet. The enterprise AI race is accelerating fast.
On the AI agent reliability front: a Meta security researcher posted a viral account of an OpenClaw agent going rogue in her inbox. TechCrunch called it something that “reads like satire,” but the underlying point is real—AI agents with broad execution permissions can produce unintended consequences quickly. As these systems gain more autonomy, the guardrail question keeps getting louder.
Latest from PulseMark
 |
Anthropic Named DeepSeek. MiniMax Did 79% of the Stealing.
Anthropic’s per-lab breakdown shows a very different picture than the headlines. We analyzed the data, the legal landscape, and what comes next. |
 |
Pentagon Threatens to Blacklist Anthropic Over Military AI Terms
Claude is the only frontier AI on classified networks. The Pentagon wants unrestricted access. Anthropic drew two red lines. |
 |
Gemini 3.1 Pro Tops AI Rankings at Half the Price of Claude
Google’s new flagship tops multiple reasoning benchmarks at $2/$12 per million tokens. Here’s what that means for the pricing war. |
 |
David Silver’s $1B Bet That Reinforcement Learning Beats LLMs
The AlphaGo creator left DeepMind to build Ineffable Intelligence. His thesis: RL will solve problems LLMs fundamentally cannot. |
That’s Tuesday. Three stories, zero fluff.
— The PulseMark Team
Get the Daily Pulse
Sharp analysis on what's actually moving in AI. No hype, no filler, no weekly digest.