@gerardsans

The Agentic AI Liability Gap: When Things Go Wrong AI Labs Blame You

4h ago · 14 min read · Anthropic markets Claude Code as an "autonomous coding agent" that "understands, plans, and executes complex tasks." Open the source code documentation and you’ll find a different story: it’s classifi

Axiom’s State of Agentic AI Q1-26: Architecture Shortcomings and Subsidised Costs

5d ago · 16 min read · Audience: CIOs, Technical Leaders, Investment Decision-MakersClassification: Public analysis based on peer-reviewed research Executive Summary The “agentic AI” narrative rests on two foundations—both

The Trillion Dollar AI Secret: Why Claude Isn't the AI System

Apr 18 · 9 min read · There is a narrative currently dominating conversations about artificial intelligence that deserves serious interrogation. It is the narrative that conflates interface with system, character with inte

OOndřej and 1 more commented

2

The Uncomfortable Truth of AI Safety: Projecting Human traits onto LLMs Is Not Science

Apr 14 · 14 min read · In February 2026, Brazilian philosopher Mariana Lins Costa published The Ghost in the Grammar: Methodological Anthropomorphism in AI Safety Evaluations (arXiv:2603.13255). With surgical precision, she

The Developing Brain vs. Unsupervised Chatbot Access

Apr 10 · 12 min read · In June 2025, OpenAI's safety team flagged an 18-year-old's ChatGPT account for "scenarios involving gun violence." Twelve employees debated whether to notify police. Leadership decided the posts didn