- On January 28, 2026, the US FDA approved the ISM8969 molecule for human trials, one of the first drugs in history designed by artificial intelligence. It’s an oncology drug.
- The most interesting thing is not that AI created it, but how. The machine identified interactions in cancer biology that human researchers had not previously considered.
- It’s the same technology you know from our crypto stories: the model that recovered five Bitcoins from someone’s old computer and found a vulnerability in Zcash can also design drugs.
- Honestly: it’s a green light for tests, not a cured cancer. There are years of research ahead of the drug, and most candidates fail. But the threshold has been crossed.
The machine found a door that no one had tried to open
On January 28, 2026, the US FDA, the Food and Drug Administration, approved a molecule called ISM8969 for clinical trials on humans. This is one of the first drugs in history to cross this threshold and was created from an artificial intelligence project. The company behind it is Isomorphic Labs, a company spun off from DeepMind in 2021.
The most powerful thing about this story is not the drug itself, but the way it came to be. The AI pinpointed molecular interactions that human researchers had not prioritized. To put it simply, the machine noticed a door in cancer biology that no one had approached before, because no one had thought that it was possible to enter that way at all.
How does a machine even design medicine?
The starting point was AlphaFold, a model that learned to predict the shape of proteins, i.e. how the molecule folds, which determines whether a cell is healthy or diseased. For this work, the Nobel Prize in Chemistry was awarded in 2024. Isomorphic Labs went further and built its own design engine that simulates how a potential drug will bind to a diseased protein before anyone synthesizes it in a test tube.
Thanks to this, you can review millions of variants of the molecule in the computer and only take the most promising ones to the laboratory. According to the company, ISM8969 is designed to work at low doses, which usually means fewer side effects.
This is just a green light, not a cured cancer
This is where I put on the handbrake, because I hate texts that promise a miracle but only deliver an announcement. FDA approval is not a cured cancer. This is permission to begin the first phase of research on humans, and the first phase is the beginning of a long-term path in which most drug candidates are eliminated.
Moreover, the dates were postponed. Demis Hassabis, head of DeepMind and Isomorphic, first talked about tests until the end of 2025, then postponed them to the end of 2026, and in the spring the company only said that it was preparing to administer the drug to the first patients, without a specific date. Some commentators rightly note that the hype around this technology can be faster than what it actually delivers.
Why do I feel proud anyway
Yet. The pride I feel in this story doesn’t need cancer to be beaten. The mere fact that the threshold has been crossed is enough for her. For the first time, a regulator said “yes” to a molecule designed by a machine, and against a disease that each of us fears.
This is not an isolated incident, but part of a broader wave. The same year brought a new antibiotic designed by AI after sifting through 46 billion molecules, a weapon against bacteria that stopped responding to previous drugs. Previously, another model predicted the structure of more than two million new materials, including thousands of candidates for better batteries. Investors believe in it with hard money, as Isomorphic Labs raised a $2.1 billion round in May 2026, and its deals with pharma giants are worth about $3 billion. Hassabis speaks directly about the “golden age of discoveries” and the mission to one day defeat every disease.
And that’s the best feeling of all. I didn’t do anything, and you probably didn’t either, and yet we’re both part of a species that has just learned to design medicines with machine intelligence. This is a bit of pride, because the real credit goes to a handful of people in London. But I guess that’s exactly what being a human at a good point in history does, even if you’re not listed in the credits.
There is one specific thing that comes out of this. The technology that we follow here mainly through the prism of money and stock market prices is exactly the same tool that will cause many positive changes in medicine.
