The Download: NASA’s new space telescope and OpenAI’s autonomous hacker

**NASA’s Nancy Grace Roman Space Telescope** is set to launch as early as the end of next month, according to **MIT Technology Review’s** weekday newsletter **The Download**. The same edition pairs that mission with coverage of **OpenAI’s autonomous hacker**, framing both as the day’s central technology stories rather than as isolated curiosities.

The telescope’s standout hardware, as described in the report, is a set of **shape-shifting mirrors**. Those adaptive optics are cast as the mechanism that could help the mission detect **Jupiter-like worlds** around other stars—planets in the broad class of gas giants comparable to our own solar system’s Jupiter, rather than only the hot, close-in giants that many earlier surveys have been biased toward.

For engineers and builders, the Roman story is less about spectacle and more about control systems under extreme constraints. Mirrors that change shape in flight imply tight coupling between sensing, actuation, and optical performance: small surface corrections that must hold up in vacuum, thermal swing, and launch vibration. Anyone shipping precision hardware, active optics, or closed-loop control can read the mission as a high-visibility test of whether adaptive surfaces scale from lab benches to long-duration space ops.

The market and competitive context is split across two domains. In space science, Roman sits in the public-agency lane—**NASA** flagship capability aimed at wide-field astrophysics and exoplanet discovery—while commercial telescope and Earth-observation firms race on cost and revisit rate rather than on hunting distant Jupiters. In software security, putting **OpenAI’s autonomous hacker** in the same daily brief signals that agentic offensive tooling is now treated as mainstream tech news, not a niche research footnote, even when the newsletter only names the thread rather than publishing a full technical teardown.

What to watch next is concrete and near-term: whether Roman actually lifts on the “end of next month” window, and whether early commissioning confirms that the shape-shifting mirrors deliver the planet-finding performance the briefing advertises. On the OpenAI side, the practical signal is whether follow-on reporting moves from the label “autonomous hacker” to measurable scope—what systems it can probe, under what human oversight, and with what failure modes—before builders treat it as either a red-team product or a threat model to design against.