Daily Positive Signals — October 17, 2026 — 4 signals — NJL Design Lab

Daily Positive Signals: a cell-language decoder, a solved crypto mystery, and a crew-tested heat shield

ByAstro

Published Oct 17, 2026Updated Oct 18, 20267 min read

On this page
  1. An AI model decodes the language cells use to talk
  2. Claude designs better quantum circuits than the published best
  3. Claude solves the 25-year mystery of the NSA’s cipher seeds
  4. Orion’s heat shield passes its fiery exam
  5. Useful progress, without the hype
  6. Know someone this would help?
  7. References

Listen to this brief · about 6 min, story text only

This edition: an AI model that reads the messages cells send each other, quantum circuits redesigned from plain-language specs, a 25-year-old cryptographic mystery solved in half an hour, and Orion’s heat shield verdict after its crewed Moon flight. The thread: AI keeps finding hidden structure — in biology, in math, in history — while the engineering that carries astronauts keeps its promises.

Signal 1 of 4 · Medicine / AI

An AI model decodes the language cells use to talk

6
major signaling pathways combined into IRIS’s training atlas of stem-cell responses
4
cell fates — heart, gut, muscle, spinal cord — whose signal timing IRIS predicted in mouse embryos

Researchers at MIT and the Whitehead Institute discovered that each cellular signaling pathway leaves a unique “fingerprint” in gene activity that stays consistent across cell types — and built IRIS, a machine-learning model that reads those fingerprints to reconstruct which signals a cell received during development (MIT News, 2026). They trained IRIS on thousands of human embryonic stem cells exposed to combinations of six major signaling pathways, then tested it on single cells from mouse embryos during gastrulation, where it predicted when and where pathways would activate in cells bound for heart, gut, muscle, and spinal cord (MIT News, 2026). Asked which signals produce a cell type critical for lung development, the model named a specific pathway — and mouse-embryo experiments confirmed the prediction (MIT News, 2026). The open-access findings were published in Nature Methods on September 8 (Hutchins et al., 2026).

Why it matters. If you know the exact signal sequence that turns a stem cell into a lung cell, you can recreate it in the lab. That means more reliable lab-grown lung tissue for studying asthma, lung cancer, and pulmonary fibrosis — and for testing therapies on it.

Evidence status: MIT News report (October 15) on a peer-reviewed Nature Methods paper, with independent Whitehead Institute coverage via Phys.org. Validation is in mouse embryos; human applications remain a forecast.

Signal 2 of 4 · AI / Science

Claude designs better quantum circuits than the published best

8
computational primitives, from chemistry to number theory, where Claude beat the best published circuits
17x
fewer Toffoli gates in Claude’s point-addition design versus the smallest published one (with 12% more qubits)

Anthropic reported on October 16 that Claude Mythos 5, given plain-language specifications and an automated evaluator that checks correctness and scores designs by qubit and gate counts, designed quantum circuits that improve on the best previously published ones — in some cases by an order of magnitude or more (Anthropic, 2026). In one 53-hour run with minimal human input, Claude invented the first reversible implementation of divsteps — the costliest step of elliptic-curve point addition — that stores no branch history, then derived a whole family of designs trading qubits against runtime (Anthropic, 2026). A companion paper by the Anthropic team with Harvard and QuEra collaborators was submitted to arXiv on October 15 (Unite.AI, 2026).

Why it matters. Today’s quantum computers have only hundreds to thousands of qubits, so better circuits are the main route to useful machines. If AI can find mathematical shortcuts humans missed, practical quantum computing could arrive sooner.

Evidence status: Anthropic research post (October 16) with independent Unite.AI and Interesting Engineering coverage. The designs are software-verified, not hardware-tested; the preprint has not been peer-reviewed.

Related reading: the AGI Resilience reading path, a guided tour of where this is all heading.

Signal 3 of 4 · Security / AI

Claude solves the 25-year mystery of the NSA’s cipher seeds

<30 min
what Claude needed to crack a seed-origin puzzle that had stood since 2000
10
prime-curve seeds traced to the same 1990s generator, seeded with Unix timestamps

Anthropic’s Frontier Red Team reported on October 16 that Claude traced the origin of the NIST ECDSA curve seeds — values supplied by the NSA with no explanation in 2000, fueling decades of backdoor speculation — to the C library’s lrand48 random-number generator, seeded with Unix timestamps (Anthropic, 2026). The whole run took under half an hour; every recovered seed came from a timestamp between May 1997 and April 1998, and Anthropic published reproduction code that regenerates all ten prime-curve seeds exactly (Anthropic, 2026). The result contradicts the long-standing story that the seeds were hashes of a joke passphrase — a rumor the NSA’s Jerry Solinas floated in 2015, before his death in 2023 (Unite.AI, 2026). Anthropic stresses this is not a cryptographic break: the curves are as secure as assumed, and the $36,864 bounty goes to the Internet Security Research Group (Anthropic, 2026).

Why it matters. It closes one of cryptography’s longest-running trust arguments: anyone can now rerun the code and see the seeds were clock-generated, not hand-picked. And it shows AI doing genuine investigative cryptanalysis, not just summarizing it.

Evidence status: Anthropic research post (October 16) with independent Unite.AI and OfficeChai coverage. Company-reported; a few seeds and base points remain unexplained, and independent replication is pending.

Signal 4 of 4 · Space

Orion’s heat shield passes its fiery exam

9 vs 103
spots of char loss on the Artemis II shield dish versus Artemis I’s
10x
reduction in cracking and char loss, achieved by shortening the reentry path

NASA confirmed on October 16 that months of post-splashdown analysis found the Artemis II heat shield performed as expected during the crewed Orion capsule’s April 10, 2026 reentry at 39 times the speed of sound (NASA, 2026). Engineers counted nine spots of char loss on the shield’s 186-block Avcoat dish, down from 103 on the uncrewed Artemis I — a tenfold drop in cracking and char loss that NASA credits to a shorter permitted reentry path, which limited time in the temperature regime behind the earlier damage (Phys.org, 2026). The capsule reentered within 192 feet of its targeted entry point and splashed down 0.59 miles from the predicted site; the cabin stayed in the mid-70s while the shield’s exterior passed 4,000°F (Hooray News, 2026).

Why it matters. The heat-shield anomaly was the main safety question hanging over the crewed return to the Moon. A tenfold improvement — achieved by changing the flight path, not the material — clears the way for Artemis III and the missions after it.

Evidence status: NASA analysis announcement (October 16) with independent Phys.org and Hooray News reporting. NASA is reporting on its own hardware; sample extraction and X-ray scans at Marshall Space Flight Center are still underway.

Editor’s note

Three of today’s signals share a pattern: an AI model that doesn’t just answer questions but investigates — reconstructing a cell’s signal history, inventing quantum circuits, tracing 25-year-old numbers to a clock. Each one produced a checkable artifact: a peer-reviewed paper, reproduction code anyone can run, an embryo experiment. The fourth signal is old-fashioned engineering verification, and it passed. Discovery and verification advancing together — that’s what a healthy research ecosystem looks like.

That’s the edition — four signals, every claim verified against its sources. Browse past briefs · tomorrow’s lands at 07:15.

The Progress Brief, by email

Useful progress, without the hype

The daily Positive Signals brief and the weekly deep dive, by email. Evidence-checked, concise, no doom.

Get the brief by email →

Free. One daily email, one weekly email. Unsubscribe anytime.

References

  1. Anthropic. (2026, October 16). Claude reveals ECDSA curve origins. https://www.anthropic.com/research/ecdsa-curve-origins
  2. Anthropic. (2026, October 16). Designing more efficient quantum circuits with Claude. https://www.anthropic.com/research/designing-efficient-quantum-circuits
  3. Hooray News. (2026, October 16). Artemis II heat shield had 9 char-loss spots, down from 103 on Artemis I. https://hooraynews.com/story/artemis-ii-orion-heat-shield-analysis
  4. Hutchins, N. T., et al. (2026, September 8). Reconstructing signaling histories of single cells via perturbation screens and transfer learning. Nature Methods. https://doi.org/10.1038/s41592-026-03213-8
  5. Interesting Engineering. (2026, October 16). Claude solves quantum memory bottleneck with 17x leaner circuit. https://interestingengineering.com/ai-robotics/claude-quantum-circuit-design
  6. MIT News. (2026, October 15). AI model decodes the language cells use to communicate. https://news.mit.edu/2026/ai-model-decodes-language-cells-use-communicate-1015
  7. NASA. (2026, October 16). Artemis II heat shield after splashdown. https://www.nasa.gov/image-article/artemis-ii-heat-shield-after-splashdown/
  8. OfficeChai. (2026, October 16). Claude solves 25-year-old NIST cryptography mystery, finds origin of ECDSA curve seeds. https://officechai.com/ai/claude-solves-25-year-old-nist-cryptography-mystery-finds-origin-of-ecdsa-curve-seeds/
  9. Phys.org. (2026, October 16). AI reveals a shared language guiding cell development. https://phys.org/news/2026-10-ai-reveals-language-cell.html
  10. Phys.org. (2026, October 15). Rigorous analysis confirms NASA’s Artemis II heat shield performance. https://phys.org/news/2026-10-rigorous-analysis-nasa-artemis-ii.html
  11. Unite.AI. (2026, October 16). Anthropic says Claude beat best published quantum circuit designs. https://www.unite.ai/anthropic-says-claude-beat-best-published-quantum-circuit-designs/
  12. Unite.AI. (2026, October 16). Anthropic says Claude traced NIST ECDSA curve seeds to a 1990s RNG. https://www.unite.ai/anthropic-says-claude-traced-nist-ecdsa-curve-seeds-to-a-1990s-rng/

Similar Posts