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

What's happening

Anthropic has publicly warned that open-source AI models can carry higher cybersecurity risks, citing a recent report that flags the Chinese company Z.ai's powerful new model (GLM-5.3) as posing a major cyberattack threat. MIT Technology Review's 'The Download' frames this as a trade-off: open models deliver clear benefits, but a new Anthropic analysis claims those benefits may be outweighed by the risks in specific cases. The conversation pairs that open-model debate with a separate theme in the newsletter about geothermal power's promise, signaling parallel discussions on technology risk versus long-term infrastructure gains.

Why it's trending

Because Anthropic's report singles out GLM-5.3 and MIT Tech Review amplified the trade-off framing, turning the open-model risk question into a fresh, widely discussed framing today.

SignalNewly emerging, confirmed across 2 independent source types.