Chronicling the technological frontier with coverage of the breakthroughs, players, and issues shaping the future. By @singularityu.

Joined June 2008
4,304 Photos and videos
The first drug targeting the RAS protein family wasn't approved until 2021 — nearly four decades after scientists discovered its role in cancer. A new one just nearly doubled survival time in pancreatic cancer patients who had already run out of options. singularityhub.com/2026/06/1…
4
752
Cells derived from Huntington's patients that formed protein clumps survived cellular stress at a dramatically higher rate than those that didn't — the opposite of what decades of research assumed. singularityhub.com/2026/06/0…
1
890
AI capabilities are advancing faster than our ability to explain them. Microsoft's chief scientific officer and a leading AI researcher just published a formal warning in Science. 🧵
4
5
16
2,562
The authors call for stronger benchmarks, better evaluation methods, and lasting norms around responsible disclosure. The goal they articulate: AI that is more intelligible and accountable — not just more capable.
1
2
430
An MIT researcher noticed that the chemical in bathroom glass etching cream dissolves the same mineral that makes lithium-rich rock so hard to process. That observation just became a startup. singularityhub.com/2026/06/0…
2
570
For decades, a group of proteins that drive some of the deadliest cancers were simply called "undruggable." A new clinical trial result suggests that label is starting to crack. 🧵
1
3
7
1,291
In a 500-person trial of advanced pancreatic cancer patients who had already tried other treatments, daraxonrasib nearly doubled median survival versus chemotherapy — 13.2 months versus 6.6 — with less pain and milder side effects.
1
445
RAS is 1 of 3 formerly "undruggable" proteins now showing cracks. The other two — p53, known as the guardian of the genome, and MYC, which drives roughly 70% of all cancers — are being approached with new strategies. AI is accelerating all of it. singularityhub.com/2026/06/1…
1
1
410
The protein clumps in Huntington's disease have been treated as the enemy for 30 years. Drugs built to destroy them have largely failed. A new study suggests scientists may have been fighting the wrong target. 🧵
1
1
1,111
The caveat the researchers are clear about: these are early results from cell models, not patients. And the protection may be temporary — what shields neurons early in the disease could become harmful as clumps accumulate over time. Timing matters enormously.
1
368
The broader question this raises: if similar dynamics apply in Alzheimer's, ALS, or Parkinson's — where protein clumps have also been the primary treatment target — the field may need to rethink not just what it's targeting, but when. singularityhub.com/2026/06/0…
387