AI & Cybersecurity
Financial Times · www.ft.com
OpenAI launched GPT-5.4-Cyber, a cybersecurity-focused model, in response to Anthropic's Mythos demonstrating strong software bug-finding capabilities.
OpenAI details its approach to scaling trusted access for its new GPT-5.4-Cyber model, positioning it as an answer to Anthropic's Claude Mythos in the cybersecurity domain.
Simon Willison · simonwillison.net
Discusses the UK AI Safety Institute's evaluation of Claude Mythos's cyber capabilities, framing modern cybersecurity as increasingly resembling a proof-of-work challenge where AI tools race to find and fix vulnerabilities.
AI Tools & Software Development
An analysis of Claude Code's source code reveals a massive 3,167-line function and regex-based sentiment analysis, raising questions about code quality in AI-generated developer tools.
Anthropic documents Claude Code Routines, a feature that lets developers turn their best AI prompts into reusable, automated coding workflows.
Financial Times · www.ft.com
Meta's plans for an AI version of Mark Zuckerberg may inspire other executives to deploy chatbot versions of themselves for internal and external communication.
A developer reflects on moving away from Agile methodologies, arguing the framework has become counterproductive in modern software development contexts.