The Download: Open model trade-offs and geothermal power’s promise

This edition of The Download explores the ongoing debate surrounding open-source AI models, highlighting concerns from Anthropic regarding the potential security risks posed by freely available, powerful models from companies like Z.ai. While safety advocates warn of misuse by bad actors, supporters emphasize the importance of democratizing access to prevent corporate monopolies. The newsletter also covers Fervo Energy’s efforts to scale enhanced geothermal power, utilizing horizontal drilling and hydraulic fracturing to expand energy production beyond traditional sites. Fervo is increasingly leveraging AI to identify new resources and optimize operations, supported by significant investment and partnerships with tech giants like Google. Additionally, the update includes reports on cyberattacks targeting food infrastructure, Meta’s latest AI model releases, and NASA’s Perseverance rover setting new driving records on Mars, reflecting the broad intersection of artificial intelligence, climate technology, and global security.
This is a summary. Read the full article at the original source:
MIT Technology ReviewRelated stories
The half-life of AI benchmarks has halved: 19 months instead of 38
A researcher analyzed 224 language model benchmarks released by nine leading labs between 2018 and 2026 to determine their lifecycle. A benchmark is c…
Google's Gemini 4 Argon is now available in the Gemini app and AI Mode in Search
Google has officially expanded the availability of its latest AI model, Gemini 4 Argon, to a broader user base. This update integrates the new model d…
Cosine Similarity Said 0.92. The Payload Was Already Dead.
The author explores the critical gap between an LLM agent's memory retrieval and actual behavioral change, challenging the common assumption that pers…



