Appearances · El podcast de la Era Exponencial
Javier Sáez Hurtado on Era Exponencial: safety and restricted models
Javier Sáez Hurtado and Arnau Ramió close the artificial intelligence series of El podcast de la Era Exponencial with this episode, published on 6 April 2026 on the Learning Heroes channel. It runs for 63 minutes.
The conversation opens with an Anthropic model whose access was restricted for safety reasons and moves on to practical questions: what benchmarks do and do not measure, when it makes sense to run models locally, and how Cowork and the other tools fit into the Claude ecosystem.
It is one of the eight artificial intelligence episodes of the series, recorded while Javier Sáez Hurtado led the IA Heroes programme at Learning Heroes.
Topics
- AI safety
- Benchmarks
- Local models
- Claude ecosystem
| Original title | “Anthropic creó un modelo demasiado peligroso (y lo escondió)” |
|---|---|
| Part of | El podcast de la Era Exponencial (episode 8 of 8) |
| With | Arnau Ramió |
| Channel | Learning Heroes |
| Published | |
| Length | 63 min |
All episodes in the series
- Javier Sáez Hurtado on Era Exponencial: AI, work and ways of thinking
- Javier Sáez Hurtado on Era Exponencial: Gemini 3 and junior roles
- Javier Sáez Hurtado on Era Exponencial: DeepSeek and efficiency
- Javier Sáez Hurtado on Era Exponencial: Claude Cowork and agents that use the computer
- Javier Sáez Hurtado on Era Exponencial: Anthropic versus OpenAI
- Javier Sáez Hurtado on Era Exponencial: OpenAI, Anthropic and defence
- Javier Sáez Hurtado on Era Exponencial: OpenClaw and autonomous agents
- Javier Sáez Hurtado on Era Exponencial: safety and restricted models