Read. Write. Edit.
In one decade, gene editing went from Nobel-winning curiosity to FDA-approved cure. The next decade will decide what we are willing to do with the tools — and what we should refuse.
Three generations of editing
What's coming
- Epigenetic editing (CRISPRoff, Weissman 2021) — heritable expression change without DNA edit
- RNA editing (ADAR, RESCUE) — reversible, no germline footprint
- Programmable retroelements / bridge RNAs (Arc Institute, 2024)
How a guide RNA finds a single base in 3 billion
What edits exist as medicine
| Therapy | Target | Stage | Year |
|---|---|---|---|
| Casgevy (exa-cel) | Sickle-cell & β-thalassemia · BCL11A KO ex vivo | FDA, EMA, MHRA approved | 2023 |
| Lyfgenia · bb305 | Sickle-cell · gene addition (lentivirus) | FDA approved | 2023 |
| Hemgenix | Hemophilia B · AAV gene therapy | FDA approved · $3.5M list | 2022 |
| Verve VERVE-101 | HeFH · in vivo base edit of PCSK9 | Phase 1b | 2023– |
| NTLA-2001 | ATTR amyloidosis · in vivo Cas9 | Phase 3 | 2024– |
| Editas EDIT-301 | Sickle-cell · γ-globin reactivation | Phase 1/2 | 2024– |
| Beam BEAM-101 | Sickle-cell · base editing | Phase 1/2 | 2024– |
forecast By 2030 we expect 10–20 approved gene-edit therapies, mostly for monogenic diseases. Polygenic disease editing remains experimental.
The malaria question
Gene drives bias inheritance so an engineered allele spreads faster than Mendelian rates. Anopheles-suppressing drives (Target Malaria, Imperial College, Crisanti lab) crashed caged populations to extinction in Italian biocontainment by 2018. The technical question is solved. The ethical and regulatory question — release into the wild — is not.
The hard questions
- Reversibility. Daisy drives (Esvelt, MIT) are designed to die out, but unproven at scale.
- Consent. Mosquitoes don't respect borders. Whose democratic process governs?
- Ecological cascade. Anopheles isn't a keystone species, but absence-of-evidence isn't evidence-of-absence.
- Dual use. The same toolkit could target useful insects.
scenario First open-environment release likely 2027–32 in Burkina Faso or São Tomé, after Phase 4-equivalent safety review.
The He Jiankui line
In November 2018, He Jiankui announced he had edited the CCR5 gene in human embryos that were carried to term — twin girls Lulu and Nana. The announcement provoked global condemnation, He's imprisonment (released 2022), and a near-universal moratorium on germline editing in clinic.
The science: the edits were imprecise, the rationale (HIV resistance) thin, the consent forms misleading. The episode established the international norm: germline editing is technically possible, currently irresponsible, and not (yet) ethically supportable.
The five accepted reasons
- Off-target risks not fully characterized
- Mosaicism — embryo-stage edits don't distribute uniformly
- Long-term inheritance unknown — multi-generational
- Less risky alternatives (PGT-M, IVF selection) exist
- Equity/eugenic concerns at population scale
contested The line will be tested again — likely first by a wealthy parent with a Mendelian-disease child, in a permissive jurisdiction.
Beyond editing — designing genomes
Synthetic biology asks: what if we wrote genomes from scratch? Craig Venter's JCVI synthesized Mycoplasma laboratorium in 2010; Sc2.0 (Boeke et al.) wrote the entire yeast genome by 2024. The toolkit:
Programmable life, in pieces
Bongard, Levin & Blackiston (2020): "xenobots" — assemblies of frog skin and cardiac cells, evolved in silico, that swim, push pellets, and self-replicate by kinematic assembly. Not a normal organism. Not a robot. Some new category we don't yet have a word for.
Anthrobots (Levin lab, 2023) — multi-cellular structures from human tracheal cells. Show neural-tissue repair effects in vitro.
scenario Living therapeutics — engineered consortia for the gut, the wound bed, the bloodstream — are a more tractable near-term application than xenobots themselves.
Cost of reading and writing DNA
Dual-use is the central problem
Drexler in 1986 worried about gray goo. The contemporary worry is more specific: cheap DNA synthesis, available pathogen sequences, and AI-assisted protein design lower the bar to engineering pandemic-class agents. Kevin Esvelt's papers (2017, 2022) lay out the case bluntly.
- Synthesis screening. IGSC member companies screen orders against hazardous-sequence databases. Coverage is partial.
- Information hazards. Some research (gain-of-function, certain pandemic-pathogen sequences) may be unsafe to publish.
- AI uplift. Frontier labs evaluate whether models offer non-trivial uplift to bioweapon synthesis. Anthropic, OpenAI, DeepMind have biorisk evals.
- Hardware. Benchtop synthesizers will eventually be cheap enough that screening at the order step matters more than at the user step.
"The scariest thing about pandemic-grade biotechnology is how few people it takes." — Kevin Esvelt
Who is shaping this
| Jennifer Doudna | Berkeley · CRISPR co-discoverer · Innovative Genomics Institute |
| Emmanuelle Charpentier | Max Planck Berlin · co-Nobel 2020 |
| Feng Zhang | Broad · CRISPR mammalian use; bridge RNAs (Arc, 2024) |
| David Liu | Broad · base & prime editing |
| George Church | Harvard · Genome Project Write; mammoth de-extinction (Colossal) |
| Kevin Esvelt | MIT · gene drives, biosecurity, daisy drives |
| Michael Levin | Tufts · xenobots, basal cognition |
| Eric Drexler | Engines of Creation (1986); molecular nanotech ground theorist |
| Françoise Baylis | Bioethicist · germline-editing critic; Altered Inheritance (2019) |
Recommended source
Kurzgesagt · "Genetic Engineering Will Change Everything Forever"
The 2016 video that introduced CRISPR to ~30M viewers. Aged remarkably well.
youtube.com/@kurzgesagt →Lex Fridman × Jennifer Doudna
Hour-plus on origins, ethics, the Broad-Berkeley patent fight, and where the science is going.
youtube.com/@lexfridman →What stays speculative
- Polygenic enhancement. Embryo selection is shipping; effect sizes are small. Editing 100s of variants in vivo, safely, is decades away. scenario
- Mammoth / thylacine de-extinction. Colossal Biosciences has cells, edit pipelines, and elephant surrogates. First mammoth-elephant hybrid maybe 2028. scenario
- Synthetic genomes for production hosts. Sc2.0 finishes; bacterial chassis with refactored genetic codes deployed in industry. forecast
- Programmable cell therapies. CAR-T 4.0; in vivo CAR generators; logic-gated kill switches. forecast
- Designer babies, mass scale. Implausible without a major regulatory shift. fiction-adjacent
Indicators · 2026–2030
- First in-vivo base-edit therapy approval (Verve, Beam)
- Open-environment gene drive trial
- Outcome of WHO & UNESCO heritable-genome-editing reviews
- Synthesis-screening federal mandate (US Executive Order 14110 successor)
- Sc2.0 yeast complete; commercial deployment
- First polygenic embryo selection birth-cohort outcome data (5+ yr)