Longevity, genes,
and disease-modifying AI.

GRAIL Galleri MCED — FDA Molecular & Clinical Genetics Panel votes in favor (2026-09-23)

After the AdComm day covered yesterday, the panel voted 10–0 for reasonable assurance of safety, 6–4 for effectiveness, and 7–2 with 1 abstention that benefits outweigh risks for multi-cancer early detection screening in adults ≥50 (prescription methylation + NGS cfDNA; Cancer Signal Origin prediction). Votes are nonbinding; FDA still decides the PMA.

Caveat: Advisory votes ≠ approval; adjunct to (not replacement for) guideline single-cancer screening; a positive cancer signal still needs diagnostic workup; verify final FDA action when posted.

Anthropic Claude discovers ART — array-associated reverse transcriptase with CRISPR-like repeats in phage DNA (2026-09-23)

AI agents (~950) mined >200k reverse transcriptases and spotted an RT + accessory gene + evenly spaced DNA-repeat array expressed as short RNAs — an architecture reminiscent of programmable nucleic-acid systems. Early lab work only; biological function unknown.

Caveat: Discovery / pre-print stage — not a proven gene-editing tool or therapy; biological role open; do not frame as “AI cured disease” or a CRISPR replacement.

Cytotoxic CD4+ T cells drive age-associated myelopoiesis via CCL5–CCR5; maraviroc rebalances aged mice (Nature Aging)

Aged bone marrow accumulates CCL5+ CD4 cytotoxic T cells (STING-linked); HSCs/GMPs upregulate CCR5 → myeloid skew and higher neutrophil-to-lymphocyte ratio. The FDA-approved CCR5 antagonist maraviroc reduced myelopoiesis, normalized NLR, and improved several aging biomarkers/function in old mice.

Caveat: Preclinical mouse data — not a human anti-aging indication; maraviroc is approved for HIV, not geroscience; human translation unproven.

TranslAGE: which epigenetic clocks actually move after longevity interventions in humans (Nature Medicine)

Harmonized analysis of 51 longitudinal human intervention studies; reliable generation-2+ clocks (especially DunedinPACE and PCGrimAge) respond most consistently; pharmacological and lifestyle interventions strongest; disease populations often show larger effects. Trial-design guidance for clinicians and trialists — not a surrogate-endpoint approval.

Caveat: Responsiveness ≠ FDA-qualified surrogate; short-term DNAm shifts ≠ proven lifespan/healthspan benefit; competing interests noted in the paper.

Base editing of HBG1/HBG2 promoters for sickle cell disease (NEJM)

Clinical base editing at fetal-hemoglobin promoters (HBG1/HBG2) for sickle cell disease — reports of durable HbF induction and clinical remission signals in treated patients. Verify exact enrollment and safety details from the full NEJM text before practice change.

Caveat: Investigational gene therapy; serious adverse events and transplant-conditioning risks apply; not a general “cure for aging.”

Research AI ≠ clinical cure claims

Nonbinding advisory votes, AI discovery of enzyme architecture, preclinical myelopoiesis axes, epigenetic-clock responsiveness, and investigational base editing are different evidence classes. This site keeps them labeled so a headline never collapses into “AI cured aging.”