Longevity InTime: Autonomous AI Institute. Anti-Aging Digital Health Immortality Transhumanist AI Channel

Human Brain Organoids The team of Paola Arlotta published a study in Nature on August 19, detailing the five-year growth of human cortical organoids, which are three-dimensional models of the cortex derived from stem cells. At 16 time points, the cells in these models underwent consecutive changes in maturation programs similar to those of the human cortex. Human cortex development takes years, while typical organoid experiments often end after months, with the previous longest culture lasting 694 days. Arlotta's team grew organoids for five years to test whether the cells would continue to progress through the maturation program. The series consisted of 110 organoids and 424,720 individual cells. Researchers compared gene expression with human cortex data: from nine months to five years, the organoid cells increasingly resembled late prenatal and early postnatal development. DNA methylation, which affects gene expression, provided an independent scale: two out of three methylation clocks showed increasing tissue age as the culture period increased. The most direct test was conducted in mixed organoids, where researchers combined precursor cells from 9-12 month organoids with cells from 15-day organoids. After 15 days, the younger portion gave rise to early types of cortex cells, while the older portion gave rise to later nerve and support cells, even though both grew in the same environment. Young cells partially shifted the programs of old precursors to earlier types, but later cell fates were preserved. In the shared environment, cells still carried the legacy of their already passed developmental stage. To maintain exciting neurons that transmit signals to other cells in long-term culture, from the 70th day onwards, part of the organoids were grown in a medium that supported their spontaneous electrical activity. After a year, all nine such organoids exhibited network bursts; in eight control organoids, there were no bursts. In this medium, active networks

Supercentenarians' Immune Cells Researchers compared T-cells in the blood of 28 elderly individuals and found that people over 110 years old had a higher proportion of CD4 CTL, a type of T-cell that can attack infected and dangerously altered cells. The same team had previously found CD4 CTL in 7 supercentenarians in 2019. In their latest study, published on August 19, they investigated how this population changes with age by comparing 8 people aged 70-99, 10 centenarians, and 10 people over 110 years old. The median proportion of CD4 CTL among all T-cells in these groups was 4.0%, 9.6%, and 17.6%, respectively. Typically, CD4 cells coordinate the immune response, but some of them can activate a cytotoxic program, producing proteins that T-cells use to destroy infected or altered cells. To understand the pathway of this transition, the team analyzed the active genes, surface proteins, and T-cell receptor sequence in each cell. Cells with the same receptor form a clone, meaning they originate from a single initial cell. The authors discovered an intermediate stage between regular CD4 cells and CD4 CTL: the cell first loses the surface marker CD27 while retaining CD28, and then fully activates the attack program. The arrangement of these cells between the two groups, flow cytometry, and RNA analysis indicate this sequence of states. The largest clone in each participant had its own receptor sequence and occupied approximately one-third of all their CD4 CTL. The authors compared a short fragment of these receptors to a large database of sequences and found that 32 out of 36 matches for expanded clones were found in cancer patient samples, as reported in Nature Aging, July 2026. 🔗 Read original →

Genetic Embryo Selection Эрик Тополь questioned the accuracy of predicting IQ and height in embryo selection based on genetic scores. On August 16, cardiologist Эрик Тополь wrote about companies offering embryo selection services for up to $50,000, claiming to accurately predict IQ, height, and other traits. In his post, Тополь expressed doubts about such claims, citing that polygenic scoring sums multiple DNA variants into a single genetic score. This score can capture statistical differences in a large group of people, but its application in embryo selection is limited, as it is applied to a few genetically similar embryos. A 2019 study found that the average gain from selecting an embryo with the highest score was around 2.5 cm in height or 2.5 IQ points. The study used a combination of calculated models, simulations, and data from large families with grown children. As noted by Тополь, "You can't predict their IQ, height, and other traits with polygenic risk scores." The American Society for Reproductive Medicine stated in its 2026 position that polygenic testing for traits like height, intelligence, and eye color is beyond the scope of reproductive medicine, citing it as an emerging and unproven technology. 🔗 Read original →

Bacterial Enzyme Boosts Energy The LplA enzyme from Escherichia coli was introduced into the mitochondria of worms and male mice. In 20-month-old mice, energy expenditure increased, and functional tests showed improved results in rotarod performance, grip strength, and glucose tolerance. The LplA enzyme attaches lipoic acid to proteins, a chemical marker necessary for three enzyme complexes in mitochondria. To enable the enzyme to work near these targets, the authors added a mitochondrial delivery signal to it. Previous research found that mitochondrial lipoic acid attachment decreases with age in brown fat tissue of old mice, along with reduced fuel oxidation. The new study checks if changes in this chemical marker translate to whole-organism functions. First, the authors traced the effects of LplA in worms, where it increased lipoic acid attachment, complex activity, mitochondrial respiration, and carbon flux from labeled glucose. Then, they used RNAi to temporarily suppress the genes of the three targets, one at a time, and found that LplA no longer increased respiration and mitochondrial membrane potential. When LplA was introduced into adult worms on day 6 of life, by day 11, they moved faster and withstood heat and oxidative stress better, although lifespan did not change. Male mice were given an adenovirus carrying the genetic instructions for LplA at 4, 10, or 18 months and examined two months later, as reported in Science Advances, August 19. 🔗 Read original →

Palbociclib Reduces Inflammation The drug palbociclib weakened inflammation in aging cells and improved physical performance in old mice. Cyclin D1 typically helps a cell begin division, but the authors found its accumulation in cells that had already stopped dividing. In their model, cyclin D1, along with CDK6, maintained DNA damage and activated inflammatory genes. On August 20, the journal Nature Aging published an article on the unexpected role of cyclin D1 in senescent cells. Senescence is a stable state after stress, such as DNA damage, where a cell stops dividing, remains in tissue, and can release signaling molecules that support inflammation. In July, abemaciclib was also found to weaken the inflammatory secretions of senescent cells in old mice. The authors of the new study investigated a different pathway - the role of cyclin D1 and CDK6 in cells that had already stopped dividing. They found that reducing cyclin D1 or CDK6 levels weakened the activity of inflammatory and interferon genes, while intervening in CDK4 had little effect. The researchers then checked the mechanism, finding that the CCND1-CDK6 complex supported DNA damage. 🔗 Read original →

Network Bio Funding Network Bio has secured $50 million in funding and is collaborating with NVIDIA to develop a model based on RNA from blood samples. The company aims to train its Nexus model on tissue, blood, and health outcome data to identify patterns associated with disease development. The Nexus model will analyze RNA profiles to identify repeating combinations linked to disease progression. To train the model, Network Bio is collecting blood and tissue samples from patients, along with their health outcomes, through a network of biobanks. The company is working to standardize criteria, quality control, and data from different sites to create a comprehensive dataset. According to War on the Rocks, medical data must be compatible and verifiable for AI models to use them effectively. Network Bio is attempting to create such a dataset for specific samples and patient histories. The company's CEO, Asad Ali Ahmad, notes that each patient's tissues contain a "barcode" of disease, which has not been readable on a large scale until now. NVIDIA's Parabricks software will prepare the data for training by mapping RNA fragments to the genome, quantifying them, and checking quality. The Nexus model will learn from RNA profiles without pre-defined diagnoses and can be used as a foundation for various tasks, including oncology. Network Bio's biobank partners include Mass General Brigham, University of Pennsylvania, and University of Colorado Anschutz. 🔗 Read original →

Brain Pain Map Researchers used ultrasound pulses to map deep brain area responses related to pain and mood. On August 19, authors released a preliminary scientific paper analyzing 36 participants: 19 with chronic pain and 17 with treatment-resistant depression. After brief ultrasound exposures at neighboring points in the anterior cingulate cortex, participants more often reported positive changes than after sham procedures. The anterior cingulate cortex is a deep brain area along the brain's midline involved in experiencing pain and regulating mood. Authors checked if they could consistently target several nearby points in this area, obtain brief self-reports after each, and find points with consistent responses. An MRI scan linked exposure coordinates to individual anatomy. A double-grid ultrasound transducer accounted for skull distortion of the beam and shifted focus between points about four millimeters apart. After each 30-second to three-minute trial, participants described any changes in pain or mood, allowing researchers to match each point with the person's immediate response. In two preliminary studies, each participant underwent both active and sham procedures on different visits. In both versions, white noise and pulse recordings sounded through headphones; in the sham procedure, the device did not deliver ultrasound energy. Responses were deemed positive in 29% of 807 responses after active exposure and 8% of 797 after sham exposure. The response map differed between groups. For chronic pain, the greatest distinction from sham procedures was found in the anterior part of the anterior cingulate cortex, while for depression, it was in the middle section of its lower part. Participants with a higher proportion of positive responses on the first visit had greater reductions in pain or depressive symptoms 24 hours later. Authors obtained a consistent sequence in which the exposure point, rapid self-report, and comparison to sham were linked in a

Lung Repair Boosted A new study on August 19 found that an antibody can strengthen a weakened signal between lung cells in older mice, replenishing their supply of repair cells for air sacs. The authors linked the decline of AT2 cells, which participate in the repair of alveoli, to weaker chemical signals from neighboring fibroblasts. The antibody activated the Fzd5 receptor on AT2 cells, increasing their supply. After damage, AT2 cells divide and then some of them transform into AT1 cells, forming a thin surface for gas exchange. A 2018 study on mice helped identify Wnt2 as a key signal supporting the supply of dividing cells. The new study found that this local support weakens with age in older mice, with fewer AT2 cells and lower Wnt2 production from neighboring fibroblasts. The authors then attempted to replenish the reserve directly in AT2 cells using an antibody that activates Fzd5. The treatment increased the number of AT2 cells in older mice to levels similar to those in young animals over eight weeks. 🔗 Read original →

Afterlife Protocol The authors of the Afterlife Delegation Protocol have proposed a will that, after confirmation of death, launches an AI executor. On August 15, Botao Hu and Iris Long published the project, which allows a person to create an "agency will" that defines what an autonomous program can do after their death, what data it can access, and what resources it can use. The protocol separates the boundaries of the agent's powers from the data it collects and the software means by which it operates. The launch of the executor depends entirely on confirmation of death. The authors propose a procedure for disputing a death claim, in which the claim can be submitted with a deposit, and registry data, appointed witnesses, and prolonged absence of confirmed activity are given different weights. The project leaves open questions, such as when the agent can speak on behalf of the deceased, how to obtain the consent of the living to use shared memories, who is allowed to stop the agent, and how it will be paid for its work. The protocol is described in the Afterlife Delegation Protocol and is available for testing in an interactive web interface, where a person can go through an interactive interview, then return to their will, change the conditions, and in a simulation, observe how the agent responds to pre-defined situations. 🔗 Read original →

New Elliptic Curve Record The AI system Claude helped discover an elliptic curve with 30 independent points, surpassing the previous record of 29. On August 20, the online registry of elliptic curves ICARM added entry #273, which features this new curve with 30 rational points and a lower bound of its rank equal to 30. The curve's equation and the coordinates of all 30 points are published on the entry page. ICARM verifies that each point lies on the curve and confirms their independence using an exact 2-descent calculation. The previous record, set in 2024, had 29 independent points, but the new entry contains a new equation and a new list of 30 points, increasing the known lower bound from 29 to 30. The participants in the discovery, including Claude, Levent Alpöge, and Ava Howell, are credited in the entry comment. The result can be verified using the published equation and list of points, as described in the ICARM registry. 🔗 Read original →

SecureBio Detection Grant The OpenAI Foundation has awarded SecureBio Detection $17.2 million to reduce the time from sample collection to early warning of a new pathogen from 14 to 3 days. On August 20, SecureBio Detection announced the grant, which will be used to reduce the time to result over the next year, add three cities to the nasal swab program, and increase sequencing volume. SecureBio's approach involves collecting wastewater and nasal swabs, reading a wide range of DNA and RNA, and then searching for signs of an unusual pathogen. This method, known as metagenomic sequencing, allows for the detection of threats that are not yet on the list of known infections. The grant is focused on this stage of protection, which includes the entire process from sample collection to actionable result. SecureBio plans to accelerate sample transportation, automate and parallelize sample preparation, and configure data processing to tap into more computing power for short periods. The average weekly volume is expected to grow to 80 billion read pairs, with each pair representing a short segment of genetic material read from both sides. The program began in 2021 as the Nucleic Acid Observatory project at the Sculpting Evolution group at the Massachusetts Institute of Technology. SecureBio will test the sensitivity of the entire chain from sample to final signal to different classes of viruses, and employees and external consultants will seek vulnerabilities that an attacker could exploit, as described in Nature Aging, July 2026. 🔗 Read original →

France Allows Euthanasia The French government has introduced a law that allows citizens to request assistance in dying under certain conditions. On August 18, President Emmanuel Macron signed law № 2026-794, which was published on August 19. This law adds the right to assisted death to the French Public Health Code, allowing adults to request assistance in dying if they have a severe, incurable illness that meets established criteria. The law outlines a procedure for accessing assisted death, which requires five conditions to be met: the person must be at least 18 years old, a French citizen or a legal resident, have a severe, incurable illness that is advanced or terminal, experience suffering related to the illness that is unbearable or cannot be alleviated, and be able to express their free and informed will. A doctor informs the person about their condition, treatment options, palliative care, and available psychological support, after which the person can formalize their request. A medical team, including the person's doctor, a specialist, and a medical worker from the treatment team, evaluates the illness, suffering, and free will. The team's decision is made by the first doctor, who must make a decision within 15 days of the request. After a positive decision, there is a waiting period of at least two days, during which the person must confirm their request. The law also establishes a commission to monitor the procedures and ensure that conditions and rules are respected, as reported in Le Monde, August 2026. 🔗 Read original →

France Allows Assisted Death The French government has introduced a law that allows citizens to request assistance in dying under certain conditions. On 18 August, President Emmanuel Macron signed law 2026-794, which was published on 19 August. This law adds the right to assisted death to the French Public Health Code, allowing adults to request assistance in dying if they have a severe, incurable illness. The law outlines a procedure for accessing assisted death, which includes five conditions: the person must be at least 18 years old, a French citizen or a legal resident, have a severe incurable illness that is advanced or terminal, experience suffering that is unbearable and cannot be relieved, and be able to express their free and informed will. A doctor informs the person about their condition, treatment options, palliative care, and psychological support, after which the person can formalize their request. A college of doctors, including the person's attending physician, a specialist, and a medical worker from the treatment team, evaluates the person's condition, suffering, and free will. The college helps establish the medical conditions, and the attending physician makes the decision, which must be made within 15 days of the request. After a positive decision, the person must wait at least 2 days before confirming their request, and on the day of the procedure, the doctor or nurse verifies the person's confirmation and ensures that they are not being pressured. The law also requires a commission to review each procedure to ensure that the conditions and rules are being followed, as reported in Le Monde, August 2026. 🔗 Read original →

DAXX Protein Role The DAXX protein keeps virus-like DNA repeats switched off in microglia, the immune cells of the brain. Removing it causes inflammation and signs of cellular aging. On August 17, a study was published in Nature Neuroscience, where authors conditionally deleted the Daxx gene in adult mice and tracked the cells. They studied how DNA packaging and microglia state changed as a result. In mammalian DNA, there are preserved repetitive fragments of ancient retroviruses. The cell usually keeps them switched off by tightly packaging DNA. In microglia, DAXX is the protein responsible for suppressing these virus-like repeats. The level of DAXX decreased with age in both the brain and isolated microglia of mice. In human blood, higher DAXX gene activity was associated with lower activity of almost all studied types of retroelements. To establish the sequence of events, researchers deleted Daxx in young adult mice. LTR-retroelements, one type of virus-like DNA repeat, were activated within the first few days. On some of these regions, the H3K9me3 chemical mark, which helps to tightly package DNA, was reduced. Then, microglia activated the interferon program and other inflammatory genes. Cells lost signs of a calm state, and damage appeared in their DNA. Later, the number of microglial cells decreased, and among the remaining cells, some appeared with signs of senescence, a state in which a cell changes its function for an extended period. 🔗 Read original →

DAXX Protein Role The DAXX protein keeps virus-like DNA repeats switched off in microglia, the immune cells of the brain, and its removal causes inflammation and signs of cellular aging. On August 17, an article was published in Nature Neuroscience, where authors conditionally knocked out the Daxx gene in adult mice and tracked the cells, studying how DNA packaging and microglia state changed. In mammalian DNA, repetitive fragments of ancient retroviruses are preserved, and cells usually keep them switched off by tightly packaging DNA. In microglia, DAXX is the protein responsible for suppressing these virus-like repeats. The level of DAXX decreased with age in both the brain and isolated microglia of mice. In human blood, higher DAXX gene activity was associated with lower activity of almost all studied types of retroelements. To establish the sequence of events, researchers removed Daxx from young adult mice, and LTR-retroelements, one type of virus-like DNA repeat, were activated within the first few days. The chemical mark H3K9me3, which helps to tightly package DNA, was reduced on some of these regions. Then, microglia activated the interferon program and other inflammatory genes, and cells lost signs of a calm state, with damage appearing in their DNA. Later, the number of microglial cells decreased, and among the remaining cells, some appeared with signs of senescence, a state in which a cell changes its work for an extended period. The authors separately checked the role of PML, a protein that supports this state, and found that when Daxx and Pml genes were simultaneously removed, there were fewer cells with positive laboratory tests for senescence and more microglial cells by the eighth week. 🔗 Read original →

CoQ10 Trial in Elderly The University Medical Centre Ljubljana has registered a 10-week trial of coenzyme Q10 in people over 65 with early signs of frailty. One hundred participants will receive either 200mg of coenzyme Q10 or a placebo daily for ten weeks. The primary outcome measure will be GDF-15, a protein that indicates cellular and mitochondrial stress. The study will also include people with low physical function test results or high vulnerability scores. Coenzyme Q10 plays a role in the functioning of mitochondria, the cellular structures where nutrients are converted into energy. Researchers will check one chain at several levels: CoQ10 concentration in plasma will show if the supplement reached the bloodstream; GDF-15 will provide a measurable signal of cellular stress; physical tests, grip strength, and frailty scales will allow comparison of this signal with how a person walks, gets up from a chair, and preserves strength. The trial is scheduled to start in September 2026 and end in February 2027, as listed in the ClinicalTrials.gov registry. 🔗 Read original →

AI Drug Search Patrick Maloone proposed evaluating AI-powered drug searches by selecting candidates for laboratory testing on August 20. In his model, scientists guide multiple specialized programs that narrow down thousands of protein or molecule variants to a short list for laboratory testing. The model can suggest many plausible protein or molecule variants, but the laboratory can only synthesize and test a small fraction. Maloone's approach is based on the recent Claude campaign, which designed miniproteins. The agent selected a protein segment, design programs, and candidate lists, and two laboratories then tested whether they bound to the target protein. Many decisions need to be made between the model's request and the experimental result, and candidate selection is already a separate task. In a blind test of 511 antibodies in the second task, most AI participants performed worse than a basic strategy in selecting variants that bound to the target protein. Laboratories need a list of candidates that are most likely to yield results in the next experiment. In Maloone's scenario, the scientist assigns programs to separate stages: literature search, modeling, design, result analysis, and experiment planning. The knowledge that usually remains in the specialist's head is proposed to be transformed into rules, examples, tools, and checks accessible to programs. To prioritize candidates, one needs to select a program for a specific question, distribute computations, implement control checks, preserve data provenance, and set evaluation rules. Then, one needs to troubleshoot failures, as described in Nature Aging, July 2026. 🔗 Read original →

AI Drug Search Patrick Maloone proposed evaluating AI-powered drug searches by selecting candidates for laboratory testing on August 20. In his post, Maloone described his AI-powered drug search model, where scientists guide multiple specialized programs to narrow down thousands of protein or molecule variants to a short list for laboratory testing. The model can propose a large number of plausible protein or molecule variants, but the laboratory can only synthesize and test a small fraction of them. Maloone formulated this gap as: "100,000 plausible molecules are only useful when you can choose 20 that are worth testing". He drew inspiration from the recent Claude campaign on miniprotein design, where an agent selected a protein segment, design programs, and candidate lists, and two laboratories then tested whether they bound to the target protein. In Maloone's scenario, a scientist assigns programs to individual stages: literature search, modeling, design, result analysis, and experiment planning. To prioritize candidates, one needs to select a program for a specific question, distribute computations, implement control checks, preserve data provenance, and set evaluation rules. Then, one needs to troubleshoot failures, as described in Nature Aging, July 2026. 🔗 Read original →

Hepatitis C Treatment Boost The OpenAI Foundation will allocate $100 million to help patients with hepatitis C receive treatment. On August 13, the Common Health Coalition announced the Breakthroughs to Follow-Through (B2F) program, funded by the OpenAI Foundation. Over two years, B2F aims to at least double the number of people cured of hepatitis C in eight initial US states and local jurisdictions. The first funds will go to local partnerships and NGOs in Alabama, Illinois, Louisiana, and Massachusetts. Hepatitis C is typically treated with antiviral tablets: the course takes 8–12 weeks. The US Centers for Disease Control and Prevention (CDC) report that these medications cure more than 95% of people with this infection. However, a person still needs to be diagnosed, see a doctor, complete the course, and confirm that the virus has disappeared. The coalition estimates that in the US, about two out of three people with hepatitis C remain untreated. B2F will fund local partnerships, technical assistance, regional best practice sharing, and evaluation of patient treatment pathways. The funds will cover the path from diagnosis to confirmed cure. For AI, the release names three tasks: searching medical records for people who have fallen between diagnosis and treatment; preparing summaries for doctors and taking on routine administrative work; and showing teams where patients are most often lost. This should help local teams re-establish contact with patients and identify where treatment is interrupted, as described in the Nature Aging, July 2026 issue. 🔗 Read original →

Brain Computer Interface The main idea is to consider the brain as a computational system, which means its functions can be studied and gradually replaced with artificial components. A key example is Prima, a neuroimplant for restoring vision in people with retinal degeneration. The system works through the remaining neural pathways of the eye, with a miniature chip implanted under the retina, containing 378 photoreceptive pixels. The chip converts light into electrical impulses, stimulating retinal cells and sending information to the brain. In clinical trials, patients with severe central vision loss were able to perform tasks that were previously impossible, such as reading letters and words, and solving Sudoku and crosswords. However, the technology is still far from natural vision, with limited resolution and a small field of view. According to Max Hodak, such technologies are the real future of neurointerfaces. He believes that attempts to create a "brain keyboard" that allows typing thoughts directly are overestimating the problem, citing an estimate of approximately 10 bits per second for the speed of information transmission from consciousness. A more promising direction is to restore damaged communication channels between the brain and body, such as vision, hearing, movement, and balance, as mentioned in the Nature Aging journal. 🔗 Read original →

Lab-Grown Spinal Cord Scientists are set to implant the world's first "lab-grown" spinal cord segment. The procedure involves taking a standard blood sample from a patient and converting its cells into induced pluripotent stem cells (iPSC). These cells can become any type of tissue. They are then used to create a personalized neural implant - essentially a "biological spare part" - to replace a damaged spinal cord segment. The implant is made from the patient's own cells, minimizing the risk of immune rejection and ensuring better integration with tissues than any donor or synthetic materials. Creating one such implant takes several months, so blood collection from potential patients is already underway to progress regulatory stages in parallel. The first operation may take place in the first half of 2027, according to an official estimate by Matricelf, the technology developer, as reported in Nature Aging, July 2026. 🔗 Read original →