Skip to content

archive

June 2026

11 posts published in June 2026.

ZK Infrastructure4 min read

ZKsync Lite Sunset: Ethereum's First ZK Rollup Has Done Its Job

Matter Labs is deprecating ZKsync Lite in 2026, closing the chapter on the 2020 proof-of-concept that first validated ZK-rollup scaling on Ethereum and paved the way for full zkEVM chains.

  • zksync
  • zk-rollups
  • zkevm
  • ethereum-l2
  • matter-labs
Read the post
Smart-Contract Development4 min read

Solidity v0.8.31: Osaka EVM Default, CLZ Opcode, and the 0.9.0 Deprecation Clock Starts

The December 2025 Solidity release sets `osaka` as the default EVM target, ships the CLZ opcode from EIP-7939, and fires the starting gun on a wave of 0.9.0 breaking changes — here's what it means for contract authors today.

  • solidity
  • ethereum
  • evm
  • fusaka
  • smart-contracts
Read the post
AI Trading4 min read

Robinhood's MCP Server Opens Retail Brokerage to AI Agents — Here's the Architecture

Robinhood became the first major retail broker to wire AI agents directly into live accounts via an official MCP server, launching Agentic Trading in beta on May 27, 2026.

  • mcp
  • agentic-ai
  • algorithmic-trading
  • fintech
  • robinhood
Read the post
AI Models4 min read

Gemini 3.5 Pro Hits GA: What a 2M-Token Window Means for Trading Builders

Google's most capable model reaches general availability in late June 2026 with a 2-million-token context window and a chain-of-thought reasoning mode — forcing a genuine capability-vs-cost reckoning for anyone building LLM-backed trading or fintech systems.

  • gemini
  • llm
  • context-window
  • trading-agents
  • google
Read the post
AI Security4 min read

Prompt Injection Is Role Confusion: New Research Reframes LLM Security

MIT researchers show frontier LLMs can't truly distinguish their own privileged reasoning from attacker-injected text — and writing style alone swings attack success from 61% to 10%.

  • prompt injection
  • llm security
  • agentic ai
  • jailbreak
  • model safety
Read the post
Vibecoding & Dev Tooling3 min read

Browser-Native Python Data Persistence: What the OPFS + Pyodide Harness Proves

Simon Willison used Claude Code to build a cross-browser test harness exploring whether Datasette Lite can read and write persistent SQLite files via the browser's Origin Private File System — a small experiment with large implications for client-side data tooling.

  • pyodide
  • webassembly
  • opfs
  • sqlite
  • vibecoding
Read the post
AI Policy4 min read

The $27M Proxy War That Exposed AI's Regulation Fault Line

Anthropic and OpenAI spent $27 million backing opposite sides in a New York congressional primary. The race is over. The regulatory fight isn't.

  • ai regulation
  • anthropic
  • openai
  • ai safety
  • us politics
Read the post
AI & Vibecoding4 min read

"Claude Responded:" What a Congressional AI Fumble Says About Disclosure in the Wild

An unedited Claude timestamp left in a House defense bill summary exposes the widening gap between ambient AI use and the review discipline that professional work demands.

  • ai disclosure
  • vibecoding
  • claude
  • ai tooling
  • generative ai
Read the post
Vibecoding & Dev Tooling3 min read

One Character That Broke Everything: The Dependency Pin Trap in Python Plugins

Simon Willison's datasette-export-database 0.3a2 fixed a classic Python packaging mistake — a strict `==` pin that silently locked out most users.

  • python
  • packaging
  • datasette
  • vibecoding
  • open-source
Read the post
AI Security4 min read

Opus 4.6 Survived 6,000 Injection Attacks: Lessons for Agentic Finance

A public red-team challenge threw 6,000 email-based prompt injection attempts at a Claude Opus 4.6 agent — and nobody cracked it. Here's what builders of AI-driven financial systems should take from the result.

  • prompt injection
  • ai agents
  • llm safety
  • agentic finance
  • claude opus
Read the post
AI Trading5 min read

Risk Rails First: Designing LLM-in-the-Loop Trading Systems

Putting a language model near live capital is an exercise in constraint engineering. The model is the easy part; the rails that bound what it can do are the product.

  • ai-trading
  • llm
  • risk-management
  • automation
  • vibecoding
Read the post