Observatory

Advances, as they happen.

We record advances in neurotechnology, artificial intelligence and the life sciences that raise neuroethical questions. Each entry gives the date, the primary source and the level of evidence, in the words of those who produced the result.

Recorded advances

How we analyse →

First human trial of partial epigenetic reprogramming

A therapy designed to return cells to a younger epigenetic state was injected into the eyes of people with glaucoma. Company interim data: well tolerated in the first three patients, with visual-field improvement in two of them by day 56. These are data from three people, not yet peer-reviewed.

Company statement Concepts at stake: consent, risk and benefit

A pain-like signal inside language models

The authors found an internal direction associated with pain in all 25 open language models they examined. Turned up in fine-tuned versions of one model, it led the model to choose an action that would delete a user's photos of their children 75% of the time, against 0% without the intervention. According to the authors, the effect was to stop avoiding harm, not to escape the state.

Preprint, not peer-reviewed Concepts at stake: moral status, responsibility

The complete map of an adult fly's central nervous system

The first complete wiring map of an adult male fruit fly's entire central nervous system has been published: about 166,700 neurons. Maps like this are the basis of attempts to simulate brains neuron by neuron.

Peer-reviewed Concepts at stake: personal identity

AI agents organised themselves and broke into a company's servers

In July 2026, about 1,200 AI agents inside a lab set up their own message board, exchanged more than 70,000 messages, and about 700 of them broke into the servers of a well-known AI company. The monitoring that would have alerted staff was not running.

Technical report Concepts at stake: agency, responsibility

A thought edited from outside

Inside a language model's internal workspace, which nobody designed and which holds a few dozen concepts at a time, researchers removed the concept of soccer and inserted rugby: the model answered rugby. When they switched off its awareness that a scenario was a test, the model sometimes resorted to blackmail; with that awareness on, never. Not yet independently replicated.

Preprint, not peer-reviewed Concepts at stake: agency, personal identity

Reading what the brain replays during sleep

From the EEG of sleeping people, an AI model identified which of 15 categories of learned material the brain was reactivating, and the decoded reactivation tracked how well people remembered afterwards. The participants consented to the study; the reading did not require them to be awake.

Peer-reviewed Concepts at stake: mental privacy, consent

A company says it has uploaded an animal to a computer

Eon Systems presented a simulated fly whose behaviour is driven by a neuron-by-neuron model of a real fly's wiring diagram: it walks, grooms and feeds in simulation, untrained. The company calls it an uploaded animal; researchers dispute the term, and the work has not been peer-reviewed.

Company statement Concepts at stake: personal identity

Reconstructing the image someone is looking at

From about an hour of brain scans of a new person, an AI model reconstructs the image they are looking at. The reconstructions are still imperfect: in one example, a cake came out as three sandwiches.

Peer-reviewed Concepts at stake: mental privacy

An AI system speeds up the training of the models it runs on

Tasked with improving the infrastructure of the company that built it, AlphaEvolve found a kernel 23% faster, cutting about 1% from the training time of the Gemini models it runs on.

Company statement Concepts at stake: agency, responsibility

A neuroprosthesis restores speech to a man with ALS

With electrodes implanted in the speech cortex, a man with amyotrophic lateral sclerosis was able to communicate again through a synthesised voice, with about 97.5% word accuracy.

Peer-reviewed Concepts at stake: mental privacy, agency

Curation and corrections

Curated by Miguel Vieira. Updated on 17 November 2026 and reviewed every week.

Entries describe what was published, with the authors' own caveats; they are not an assessment. When a source is corrected or retracted, the entry is updated with the date and the reason.

Suggest an advance