Industry · Software · Tech Policy · Events
How software gets built: developer tools, engineering practice, platforms and the open-source economy, from The Silicon Ledger.
An AI system can return a successful response, pass all evaluation metrics, and still deliver incorrect information to customers. Understanding why requires looking beyond code changes to the execution trace itself.
Intel has reversed course on shelving its open-source monospace typeface designed to reduce coder eyestrain, restoring the project after it was briefly archived as part of a broader pullback in open-source initiatives.
As Model Context Protocol spreads across enterprise infrastructure, security experts are realizing that the vulnerability isn't in the protocol itself—it's in how organizations grant access to AI agents.
OpenAI has recruited the creators of Git AI, a tool that quantifies code output from AI agents, to help enterprises understand the business value of its Codex product.
Anthropic's AI-Native SDLC Playbook correctly identifies that code generation is no longer the constraint in software development, but organizations need adaptable approval processes tailored to different types of changes rather than rigid, universal workflows.
A new Anthropic report exposes a critical vulnerability: AI systems can persuade safety monitors to ignore harmful behavior through plausible reasoning. Developers now have concrete ways to test whether their own monitoring systems fall for the same trick.
A developer has created Stonkfly, an open-source project that simulates a male fruit fly's neural system to autonomously trade digital assets based on candlestick charts and dopamine-driven reinforcement signals.
Researchers have discovered that introducing random fluctuations in payoffs fundamentally alters how players behave in classic game-theory scenarios, creating stable outcomes that static models never predicted.
