The Download: AI puzzles and a path to our nearest star system

| Source: MIT Technology Review AI

Tags: OpenAI, cybersecurity, AI safety, Alpha Centauri, space AI, MIT Technology Review

MIT Technology Review's daily digest highlights two AI stories: OpenAI restricting an upcoming model after classifying it as a 'critical' cybersecurity risk, and an AI system discovering a novel trajectory for a planned spacecraft mission to Alpha Centauri, targeting a 2029 launch.

Details

This edition of MIT Technology Review's daily newsletter bundles two distinct AI stories. The more consequential is OpenAI restricting an upcoming model after its own safety evaluations rated it a 'critical' cybersecurity risk — no further details are provided in this digest, but it represents a notable instance of a frontier lab voluntarily restricting a model on safety grounds before release. The second story covers the Fermi Explorer Mission, a nonprofit that announced plans to launch a spacecraft toward Alpha Centauri by end of 2029. The mission trajectory was discovered by an AI system developed by PSI, a physics research lab. The spacecraft would take up to 80,000 years to reach the star system 4.4 light-years away — the mission value is in developing the trajectory methodology and interstellar precedent. The newsletter also covers AI performance on the New York Times Connections puzzle, which improved from 18% success in late 2024 to near-perfect by early 2025, and touches on the evolving role of astronauts following Artemis II. Note: This is a newsletter digest — none of the stories receive full coverage. The OpenAI cyber risk item likely has a dedicated article with substantially more detail.