Genetic Engineering and Neurotechnology

Linas Juozenas

Intelligence Unleashed · Biology & brain stimulation

Genetic engineering & neurotechnology

CRISPR can alter DNA. TMS and tDCS alter electrical conditions in neural networks without rewriting the genome. They operate on different timescales, carry different risks and sit at very different points on the path from laboratory result to routine care.

Start with the boundary

Changing a sequence is not the same as changing a signal

The phrase neurotechnology covers tools that can act at radically different layers of biology. Genome editing changes molecular instructions inside selected cells. Brain stimulation changes the electrical conditions under which networks operate. Putting both under one futuristic label makes it easy to confuse a permanent intervention with a repeatable dose—or a mouse experiment with an available treatment.

The sequence lever

Genome editing

Targets DNA in a cell. Depending on the editor, it may cut, replace, disable or rewrite a chosen sequence. The intended change can persist for the life of that cell.

Unit of action: a genomic site in selected cells
Delivery: cells edited outside the body, or cargo delivered into tissue
Reversibility: generally low once a cell has been edited
Neural status: mainly cell and animal research, with limited ocular human experience

The circuit lever

Neurostimulation

Applies magnetic, electrical, acoustic or implanted electrical energy to influence neural activity and plasticity. Effects depend on dose, target, timing and the state of the network.

Unit of action: cells, pathways or networks within an induced field
Delivery: external coil or electrodes, or an implanted device
Reversibility: the applied energy stops; biological effects may outlast a session
Clinical status: some tightly specified uses are cleared or approved

Reality map · September 2026

What has actually reached people?

The strongest way to read any breakthrough claim is to ask two questions: what tissue was changed? and what level of evidence supports the claim? A therapy for neurological symptoms may edit the liver rather than the nervous system. A CRISPR-branded intervention may target RNA rather than DNA. And “in a human” does not necessarily mean an authorized, controlled or successful clinical trial.

Established outside neurology

Ex-vivo CRISPR

CASGEVY provides a clinical benchmark: a patient’s blood-forming stem cells are removed, edited and returned. FDA first approved it in 2023 and expanded the eligible ages in 2026. It validates ex-vivo hematopoietic editing—not delivery into the brain.²

Small human study

Direct retinal DNA editing

EDIT-101 delivered CRISPR components beneath the retina in 14 people with CEP290-associated LCA10. The uncontrolled phase 1–2 study reported no dose-limiting toxicities and functional improvement on at least one measure in some participants. Retinal tissue could not be biopsied, so editing in photoreceptors was not directly quantified.³

Predominantly preclinical

Editing neuronal or glial DNA

Promising results exist in cultured cells and animal models, including 2026 mouse studies in Huntington’s disease and Dravet syndrome. These are not proof of safe, effective brain editing in people.

No established therapy

Human brain genome editing

No brain genome-editing treatment is approved, and no successful regulated trial has published efficacy. A single-patient CHD3 study in China was posted on ClinicalTrials.gov before dosing but proceeded without national regulatory review; the child died from a treatment-related complication seven days after dosing.⁴⁵

Evidence is a staircase

A result does not climb automatically

Mechanistic plausibility, a corrected cell, a healthier mouse, an early safety signal and a licensed therapy answer different questions. Each higher rung requires new evidence.

Mechanism
Can the tool make the intended molecular or physiological change under controlled conditions?
Cells and organoids
Does it work in relevant human-derived cells, and what unintended effects appear?
Animal models
Can it reach the target, alter a disease-relevant endpoint and avoid unacceptable harm?
Early human study
Is administration feasible and tolerable in a small, carefully selected group?
Controlled clinical evidence
Does it outperform an appropriate comparator on meaningful outcomes?
Authorization and practice
Has a regulator reviewed a specific product, population, indication and protocol—and do benefits hold in real care?

The genome-editing toolbox

“CRISPR” is a family of operations, not one molecular scalpel

A guide RNA can bring a CRISPR-associated protein to a chosen sequence, but what happens there depends on the system. Precision at the address does not guarantee precision in the outcome: repair pathways, cell type, editor exposure and the assay used to look for errors all matter.

Common CRISPR-associated approaches and their limits
Approach	Primary action	What it can do	Important qualification
Cas9	Cuts both DNA strands at a guided site	Disrupt a gene or enable a designed repair	Double-strand-break repair can create mixed outcomes, large deletions or rearrangements.
Cas12a	Creates a staggered double-strand DNA break	Target sites and multiplexing options that differ from Cas9	It is still a nuclease. It should not be grouped with editors that avoid programmed double-strand breaks.
Base editor	Chemically converts one DNA base into another within an editing window	Correct certain point mutations without a programmed double-strand break	Possible bystander edits, guide-dependent off-target activity and unintended on-target changes remain.
Prime editor	Copies a template encoded in a prime-editing guide RNA	Install substitutions and relatively small insertions or deletions	Efficiency and by-products vary by target. Large cargo insertion requires additional architectures and harder delivery.
Cas13	Targets RNA	Reduce or alter a transcript without changing DNA	It is a CRISPR system, but it is not genome editing. Its effect may require durable expression or repeat dosing.

The target can be correct while the outcome is not

An editor may reach the intended genomic site yet produce unintended insertions, deletions, structural variants or mixtures of edited and unedited cells. “On target” is therefore not synonymous with “as designed.”

Off-target risk has no universal percentage

Rates depend on the locus, guide, editor, dose, tissue and measurement method. A platform-wide promise such as “under 0.1%” is not interpretable without product-specific data and a stated detection limit.

One sequencing test cannot close the case

Whole-genome sequencing may miss rare or complex events. FDA guidance expects complementary, product-specific assessment of nominated off-target sites, unintended on-target outcomes and chromosomal integrity.⁶

What the neurological examples really show

Selected neurological targets by evidence stage
Target or disease	What has been demonstrated	What has not
HTT · Huntington’s disease	A 2026 dual-AAV9 base-editing study injected mouse striatum and altered an exon-13 splice acceptor rather than the disease-causing CAG repeat. At 12 months, the authors reported about 16% editing in bulk striatal tissue and 9% exon skipping—not uniform brain-wide correction.⁷	The human phase 1–2 AMT-130 trial is an RNA-lowering gene therapy, not CRISPR. No human HTT DNA-editing efficacy has been established.
SCN1A · Dravet syndrome	Variant-specific adenine base editing improved disease-relevant outcomes in a mouse model in 2026.⁸	No human SCN1A base-editing trial was identified in ClinicalTrials.gov as of 4 September 2026. ETX101 is an AAV9-delivered gene-regulation therapy that increases SCN1A transcription; it does not edit DNA.⁹
APOE · Alzheimer’s-risk biology	Prime editing from APOE4 toward APOE3 has been studied in human induced cells and mouse models.¹⁰	No human administration, disease-modifying benefit or population-level prevention strategy has been established.
MECP2 duplication	An open-label early study with an estimated six participants is evaluating a single intracerebroventricular dose of an AAV vector encoding Cas13Y to reduce excess MECP2 RNA.¹¹	Cas13 does not rewrite genomic DNA. The registry showed no posted results as of 4 September 2026.

The decisive bottleneck

An editor cannot help a cell it cannot safely reach

For the brain, delivery is not a packaging footnote. A therapeutic system must cross or bypass protective barriers, spread far enough through complex tissue, enter the right cell type, express at a useful level and then stop—without provoking an immune or toxic response.

Package

Fit the editor, guide and regulatory elements into a deliverable format.
Route

Choose systemic, intrathecal, intraventricular or direct tissue administration.
Distribute

Reach the needed region and cell population at a workable dose.
Edit

Create the intended outcome in enough cells while limiting by-products.
Follow

Monitor immune, genomic and clinical effects over an appropriately long period.

Viral vector

Adeno-associated virus

AAV can provide durable expression and has substantial gene-therapy experience. But its cargo capacity is about 4.7 kilobases, often too small for an editor, promoter and guide in one vector. Splitting a system across vectors requires the same cell to receive and reconstruct all needed parts.

StrengthEfficient delivery to certain tissues and cell types

ConstraintPayload, immunity, dose toxicity and limited redosing

Non-viral carrier

Lipid nanoparticles

LNPs can carry transient RNA and reduce the duration of editor exposure. Clinically proven formulations preferentially reach the liver, however. Designs that cross the blood–brain barrier and target particular neural cells remain largely preclinical.

StrengthTransient editor exposure and scalable manufacturing

ConstraintBrain entry, distribution and cell-specific uptake

Local administration

Direct or cerebrospinal-fluid routes

Injection into brain tissue, ventricles or the intrathecal space bypasses parts of the blood–brain barrier. It also adds procedural risk and does not guarantee even distribution across a large or anatomically dispersed disease network.

StrengthHigher exposure near an intended region

ConstraintInvasiveness, focal spread and heterogeneous editing

“Crosses the blood–brain barrier” is not a yes-or-no property of AAV9—or any vector. Performance changes with capsid, route, dose, age, species and cell type. Vectors that look powerful in rodents can show far less neuronal delivery in nonhuman primates.¹²

Why the retina is a special case

The retina is neural tissue, but it is small, compartmentalized, optically assessable and reachable through local surgery. EDIT-101 is important evidence that in-vivo editing can be administered to human neural tissue. It is not evidence that a systemic editor can traverse the adult human blood–brain barrier or correct a diffuse brain disorder.

Why post-mitotic neurons are hard

Mature neurons rarely divide, so repair strategies that depend on homology-directed repair are inefficient. Base and prime editors can work without that pathway, but they are large molecular systems, can generate their own unintended outcomes and still face the same delivery problem.

A credible safety package must look in more than one place

Genome

Intended edit, unintended edits at the target, off-target sites, structural variants, mosaicism and the sensitivity of each assay.

Body

Biodistribution, vector shedding, immune response, organ toxicity, dose relationship and effects beyond the intended cell population.

Time

Durability, delayed neurological effects, tumor risk where relevant and a long-term follow-up plan proportionate to an irreversible intervention.

Non-invasive brain stimulation

The dose is a protocol, not merely a current or a coil

TMS and tDCS do not “switch on intelligence.” They perturb excitable tissue under a defined set of conditions. The clinically meaningful unit is the complete protocol: device, waveform, intensity, target, session length, number and spacing of sessions, population, co-treatments and outcome.

Electromagnetic induction

Transcranial magnetic stimulation

A rapidly changing current in a coil creates a magnetic field, which induces an electric field in superficial cortical tissue. A single pulse can probe physiology; repeated trains can alter network excitability and plasticity.

rTMSRepeated pulses at a specified frequency and intensity

TBSBursts patterned at theta-burst timing, delivered intermittently or continuously

Deep TMSBroader coil geometries that trade some focality for a different field distribution

Weak direct current

Transcranial direct-current stimulation

Electrodes on the scalp pass a low-intensity current through the head. The resulting intracranial field is weak and diffuse; it changes the probability that neurons will fire rather than directly forcing a synchronized discharge.

MontageElectrode size, placement and return path shape the field

DoseCurrent, duration, ramping, repetition and spacing all matter

StateTask, baseline activity, anatomy, medication and prior exposure can change the effect

Why similar-looking protocols can produce different results

Where is the field strongest?

Scalp landmarks are approximations. Individual anatomy, coil orientation and electrode placement change what tissue receives the dose.
What state is the network in?

Sleep, attention, medication, disease, recent activity and concurrent training can influence the same nominal stimulation.
What outcome is measured?

A short-lived laboratory change, a symptom-scale improvement and durable functional recovery are not interchangeable endpoints.
What is the comparator?

Blinding is difficult when participants feel scalp sensations. Credible trials need a sham condition that addresses expectation and unblinding.

Clinical and regulatory status

A regulator authorizes a product for a use—not a technology for everything

“FDA cleared” and “FDA approved” are not decorative synonyms. Many TMS devices reach the U.S. market through De Novo classification or 510(k) clearance, with indication-specific controls. In December 2025, one tDCS system received premarket approval. None of those decisions validates a homemade montage, a different headset or an off-label claim of general cognitive enhancement.

Selected U.S. clinical status as of 4 September 2026
Technology and use	Status	What the status does—and does not—mean
TMS for major depressive disorder	Cleared products	Multiple devices are cleared under specific labels, generally after nonresponse to prior medication.¹³ One FDA record covers reducing comorbid anxiety symptoms in adults treated for MDD—not a standalone anxiety-disorder indication.¹⁴ In 2024, one system added adjunctive treatment for ages 15–21.¹⁵
TMS for obsessive-compulsive disorder	Cleared products	Certain deep-TMS systems and protocols are cleared for adults. Coil, target, symptom provocation, schedule and device labeling matter.
TMS for smoking cessation	Cleared product	A specified deep-TMS device is cleared as an aid in short-term smoking cessation for adults, alongside a defined protocol—not as a universal addiction treatment.¹⁶
TMS for post-traumatic stress disorder	Cleared product · 2026	The MeRT System was cleared in June 2026 as an adjunctive treatment for adults. FDA noted that the pivotal trial tested the whole system and did not isolate whether its EEG-personalized frequency adds benefit over fixed-frequency TMS.¹⁷
TMS for migraine	Cleared devices	Some prescription migraine devices use single-pulse TMS for acute and preventive treatment. Their authorization should not be attributed automatically to repetitive TMS used in psychiatry.¹⁸
Flow FL-100 tDCS for major depressive disorder	PMA approved	A prescription, at-home system for adults with a current moderate-to-severe major depressive episode, as monotherapy or alongside other antidepressant treatment; the label excludes treatment-refractory patients.¹⁹
Generic consumer tDCS for focus, memory or wellness	Not established	One product’s approval cannot be transferred to another device, placement, dose, user group or claim. “Wellness,” “FDA registered” and “FDA approved” mean different things.

TMS

What a treatment course looks like

Outpatient TMS commonly involves repeated sessions over several weeks, though accelerated and theta-burst schedules can differ. The patient remains awake. Ear protection, coil positioning and motor-threshold or other dosing procedures are part of treatment—not optional setup.

Common effects include scalp discomfort, muscle twitching and headache. A seizure is uncommon but possible, so clinicians screen for seizure risk, medications and metal or implanted devices near the head. Expert safety guidance emphasizes that risk depends on protocol and patient factors.²⁰

tDCS

What the 2025 approval changed

The December 2025 FDA decision applies to one prescription at-home tDCS system for a defined depression population. In its 174-participant, ten-week pivotal trial, the mean HDRS-17 depression score fell 9.4 points with active treatment and 7.1 with sham, a 2.3-point between-group difference. Response was reported in 54.4% versus 26.9%, and remission in 44.9% versus 21.8%.

The FDA review also recorded meaningful uncertainty, including risks of unblinding, no prespecified threshold for a clinically meaningful HDRS-17 change and tension with parts of the prior literature. Approval supports the labeled benefit–risk decision; it does not erase the study’s limitations.²¹

Most reported effects were local and temporary, but two first-degree burns occurred at 2 mA after electrode pads dried. Burn risk is therefore not defined by crossing a simple current threshold; electrode condition, contact and skin checks matter.

Before treatment, ask for the label-level answer

Which exact device?

Manufacturer and model—not just “TMS” or “tDCS.”
Which indication and population?

Diagnosis, severity, age range, prior treatment and monotherapy or adjunctive use.
Which protocol?

Target, waveform, dose, course length and any required behavioral task or co-treatment.
Who supervises and follows up?

Screening, adverse-event plan, response measures and what happens if symptoms worsen.

Treatment is not enhancement

Small task effects are not “more intelligence”

Brain-stimulation studies in healthy participants can detect changes in a particular task under a particular protocol. That is scientifically interesting. It does not establish a durable, transferable increase in general intelligence, creativity or learning capacity.

Effect size

A 2024 meta-analysis of offline high-frequency rTMS in healthy people found statistically detectable but small average effects on accuracy and reaction time. The estimates were roughly one tenth of a standard deviation—not a cognitive transformation.²³

Consistency

Umbrella-level evidence for prefrontal tDCS and cognition has been conflicting. The review rated the included meta-analyses low or critically low in methodological quality, and most significant effects rested on low- or very-low-certainty evidence.²⁴

Transfer

Improvement on the trained outcome does not necessarily generalize to school, work or daily life. A study should test far transfer and persistence rather than infer them from immediate performance.

The consumer-facing conclusion is narrow: no non-invasive brain-stimulation device is FDA cleared or approved for general cognitive enhancement in healthy users, and the research does not support a dependable “upgrade” to intelligence.

Personalization

Research signal, not a prescribing tool

Researchers study anatomy, connectivity, symptoms and genes such as BDNF Val66Met as possible sources of response variability. Human findings are mixed, and a 75-patient naturalistic depression study reported no genotype difference in rTMS response or remission. That one cohort is not dispositive, but it reinforces that genotype-guided TMS is not routine clinical practice.²⁵

Where the fields may meet

The components are advancing separately; the full stack does not yet exist

It is reasonable to imagine genetics helping select a therapy, sensors adjusting stimulation in real time, or gene delivery making a circuit light-sensitive. It is not reasonable to present those separate lines of work as one integrated clinical platform. The right description is an evidence map, not a countdown.

What is real, what remains experimental and what is still hypothetical
Convergence idea	Best evidence today	Boundary
EEG-triggered stimulation	A small human experiment detected ongoing sleep spindles and delivered feedback-controlled 12-Hz tACS, reporting increased spindle activity and better motor-memory consolidation.²⁶	Laboratory proof-of-concept in 16 participants, not an established memory treatment and not genome editing.
Genetics-informed stimulation	Candidate variants such as BDNF Val66Met are studied as moderators of physiological and clinical response.	Findings have not produced a validated test that chooses an rTMS protocol for an individual patient.
Optogenetics	In one participant with retinitis pigmentosa, a retinal study combined AAV-delivered ChrimsonR expression with light-stimulating goggles; animal studies have explored deeper targets.²⁷	CRISPR is not required. The human evidence is retinal; it does not establish deep-brain optogenetic treatment and did not use CRISPR.
Low-intensity focused ultrasound	A sham-controlled human experiment showed target engagement in the amygdala.²⁸	Neuromodulation efficacy remains investigational. It is distinct from MR-guided focused-ultrasound ablation used for certain movement-disorder procedures.
Bidirectional implanted BCI	In a single-participant study, motor-cortex recording controlled a robotic arm while somatosensory stimulation supplied tactile feedback and improved performance.²⁹	This is genuine recording-plus-stimulation evidence. It did not release an editor or change the participant’s DNA.
BCI-triggered genome editing	No integrated animal or human demonstration was identified.	Recording, stimulation, microfluidics and genome-editor delivery exist as separate research areas. Current evidence does not support assigning a clinical deployment date.

A useful forecasting rule

Count the unsolved interfaces

A concept that needs sensing, classification, a safe trigger, a payload reservoir, tissue-specific release, cellular uptake, accurate editing and long-term control is not one breakthrough away. Every interface adds a failure mode, validation burden and regulatory question.

Ethics follows the actual intervention

Permanence, power and data change the moral landscape

A voluntary, time-limited clinical stimulation session is ethically different from an irreversible edit delivered into a child’s brain—or from monitoring workers’ neural signals. Governance should follow the intervention’s real properties, not its marketing category.

When an intervention may persist

Genome-editing questions

Is the disease serious enough to justify an irreversible risk?
Can the intended cells be reached without exposing unrelated tissue?
How will rare or delayed effects that are not heritable but may be lifelong be detected?
What care is guaranteed if a sponsor, institution or trial closes?
Can a child assent, and is waiting for adult consent medically possible?

When a system senses or stimulates

Neurotechnology questions

Is consent genuinely voluntary in school, work, military, prison or care settings?
What neural data are collected, inferred, retained and shared?
Can the user pause treatment, delete data and leave without penalty?
Who can access raw neural signals and inferences about mood, attention or preference—and how can errors be corrected?
Will refusing enhancement reduce access to education or employment?

Six principles that survive the hype cycle

Proportionality

Risk, permanence and uncertainty should be proportionate to the condition’s severity and the likelihood of meaningful benefit.

Iterative consent

Consent is a continuing process, especially when evidence changes or a device begins collecting new categories of data.

Agency

A person should be able to refuse monitoring or enhancement without losing ordinary rights, services, schooling or work.

Privacy by design

Collect the least neural data needed, limit purpose and retention, secure access and make secondary use opt-in.

Equity

Research populations, access, price and benefit sharing should not turn public risk-taking into private advantage for a narrow group.

Disability justice

Treatment choice must coexist with respect for disabled lives. A molecular “correction” narrative should not define a person as a defect.

UNESCO’s 2025 Recommendation on the Ethics of Neurotechnology emphasizes autonomy, mental integrity, neural-data governance, child protection, fairness and scrutiny of workplace or educational use.³⁰ The OECD’s responsible-innovation framework likewise calls for stewardship, safety, inclusion and the capacity to anticipate misuse.³¹ These are influential international standards, not self-executing laws in every country.

Regulation also depends on intended purpose and jurisdiction. In the European Union, non-medical equipment that applies electrical, magnetic or electromagnetic fields through the cranium for neuronal stimulation falls under Annex XVI and is reclassified as Class III; medical-purpose devices follow separate MDR rules.³³ In the United States, FDA’s general-wellness policy for low-risk products does not confer device clearance or approval.³⁴

A claim-reading toolkit

Eight questions for the next “brain breakthrough”

The fastest protection against hype is not cynicism. It is specificity. These questions work for a paper, a press release, a clinic page, an investor deck or a consumer headset.

What was changed?

DNA sequence, RNA abundance, protein expression, electrical activity, imaging signal, symptom score or task performance?
Where did it happen?

Dish, organoid, mouse, nonhuman primate, retina, liver, cerebrospinal fluid or human brain? Disease symptoms do not identify the edited tissue.
Which evidence rung?

Mechanistic study, animal proof-of-concept, first-in-human safety, uncontrolled signal, randomized trial, regulatory decision or routine-care outcome?
Compared with what?

Untreated cells, healthy animals, sham stimulation, standard care or an active comparator? Was allocation concealed, and did blinding actually work?
How large and durable?

Look for absolute change, confidence intervals, participant count, attrition, clinical importance, follow-up length and whether the effect generalized beyond one task.
What else changed?

For editing: by-products, off-target sites, biodistribution and immunity. For stimulation: adverse events, mood worsening, unblinding and concurrent treatment.
What exactly did a regulator decide?

Name the device or product, pathway, indication, population, contraindications and date. “FDA registered” is not a therapeutic authorization.
Who has replicated it?

Separate peer-reviewed evidence from a company release. Check preregistration, protocol changes, conflicts of interest, corrections and independent replication.

What to watch next

Milestones matter more than dates

Forecasts age badly in fast-moving fields. Observable milestones give a better view of progress because they identify the bottleneck a new result actually removes.

Genome delivery

Translation across species

Look for systemic or minimally invasive vectors that reproduce meaningful brain distribution and cell specificity in nonhuman primates—not only rodents—at tolerable doses.

Genome safety

Product-specific error maps

Watch for complementary assays of unintended on-target events, off-target edits, immune response and biodistribution, plus transparent long-term follow-up.

Stimulation efficacy

Independent head-to-head trials

Personalized targets, accelerated schedules and closed-loop timing need comparisons that isolate the proposed innovation—not trials of a bundle whose active ingredient remains unknown.

Home treatment

Post-market reality

For prescription home tDCS, watch adherence, skin injury, mood switching, clinician workload, cybersecurity and effectiveness outside a pivotal trial.

Measurement

Outcomes people can feel

A biomarker can show mechanism or target engagement, but it does not establish clinical benefit unless it reliably predicts durable gains in symptoms, function or quality of life. Look for outcomes chosen before results were known.

Governance

Rights that work in practice

Track enforceable rules for neural data, children, work and education; trial continuity when a sponsor exits; affordable access; and meaningful remedies when systems are misused.

The most plausible near-term future is not a device that rewrites intelligence on demand. It is narrower: better targeting, shorter or more adaptive stimulation protocols, improved delivery to selected tissues, and more disciplined evidence about who benefits.

Frequently asked questions

Short answers to the biggest questions

Is any CRISPR treatment approved for the brain?

No. CRISPR has approved ex-vivo uses in blood disorders, and in-vivo editing has reached small human studies in tissues such as the retina and liver. No genome-editing treatment is approved to alter neuronal or glial DNA in the human brain.

Did EDIT-101 prove that CRISPR can edit the human brain?

No. It was an important direct retinal study. The retina is neural tissue but differs greatly from the brain in access, scale and compartmentalization. Because researchers could not biopsy participants’ retinas, the trial also could not directly measure the genomic edit in treated photoreceptors.

Can CRISPR be used to increase intelligence?

There is no demonstrated, safe route to do so. Intelligence is a broad, context-dependent set of abilities influenced by many genes, development, health, education and environment. Editing one variant would not amount to a controllable intelligence setting, and irreversible neural intervention would carry profound uncertainty.

Are TMS and tDCS the same?

No. TMS uses electromagnetic induction to create a comparatively strong, time-varying electric field in cortex. tDCS sends a weak direct current between scalp electrodes to bias excitability. Their field shapes, doses, mechanisms, risks and regulatory evidence differ.

Does TMS cure depression?

TMS is an evidence-based treatment option for appropriately selected people, but it is not a guaranteed or permanent cure. Some patients respond or remit, others do not, and maintenance or further treatment may be needed. Outcomes depend on the device, protocol, diagnosis and individual clinical context.

Can tDCS make a healthy person smarter?

No dependable general-intelligence benefit has been established. Research effects are often small, inconsistent and tied to a particular task and protocol. An approved depression device is not an approved cognitive-enhancement device.

Is at-home tDCS safe?

One U.S. prescription system has an approved, clinician-managed home protocol for a narrow depression population. That finding cannot be generalized to all headsets or DIY use. Electrode condition, placement, skin integrity, dose, contraindications, medication and monitoring all matter; burns can occur even at a commonly studied current if contact is poor.

Can a gene test choose the best TMS protocol?

Not in routine care. Variants such as BDNF Val66Met are active research topics, but findings are inconsistent and have not produced a validated clinical selection algorithm.

Will a brain–computer interface eventually trigger CRISPR editing?

That is hypothetical. Closed-loop sensing and stimulation, implantable drug delivery and genome editing each exist as research areas, but no integrated system has demonstrated BCI-triggered genome-editor delivery in animals or people. Assigning a clinical date would be speculation.

Would editing neurons change a patient’s descendants?

Not by intent. Neurons are somatic cells, and a localized neural edit is designed to affect the treated person. Heritable editing involves gametes or embryos. Researchers must still study whether a delivery system reaches tissue beyond its intended target.

The durable conclusion

Power requires sharper distinctions

Genome editing and neurostimulation are both becoming more precise, but precision is not a single property. A guide RNA can be sequence-specific while its delivery is diffuse. A TMS coil can be positioned precisely while the relevant network response varies. A statistically significant result can still be clinically modest.

The mature view is neither utopian nor dismissive. CRISPR has entered clinical practice for selected blood disorders, while neural applications remain predominantly preclinical and safe delivery and irreversible-risk control remain formidable. TMS is already useful for selected patients, and a prescription tDCS system now has a narrowly defined U.S. approval, while consumer enhancement claims run far ahead of evidence. The work ahead is to turn technical possibility into outcomes that are reproducible, proportionate, accessible and chosen freely.

Evidence review: clinical, regulatory and ethics sources checked through 4 September 2026. Regulatory decisions are jurisdiction-, product-, population- and indication-specific and may change. This educational overview is not medical advice.

Sources and further reading

Selected evidence, official records, guidance and reporting

Sources are listed in first-use order. Trial status and regulatory scope were checked on 4 September 2026.

ClinicalTrials.gov: AMT-130 in early manifest Huntington’s disease (NCT04120493) (opens in a new tab). Registry record identifying the investigational AAV5 microRNA gene therapy.
U.S. Food and Drug Administration: CASGEVY (opens in a new tab). Product information for the ex-vivo CRISPR-edited hematopoietic stem-cell therapy.
Pierce et al. In vivo CRISPR gene editing for inherited retinal disease. New England Journal of Medicine (2024) (opens in a new tab). Phase 1–2 EDIT-101 report.
Science: Death of girl in Chinese gene-editing trial was never made public (2026) (opens in a new tab). Investigation of the CHD3 base-editing case, oversight pathway and delayed disclosure of the treatment-related fatal event.
ClinicalTrials.gov: single-participant CHD3 base-editing study (NCT06860672) (opens in a new tab). Official registry record posted before dosing.
FDA: Human Gene Therapy Products Incorporating Human Genome Editing—Guidance for Industry (2024) (opens in a new tab). Product-specific assessment of off-target, unintended on-target and long-term risks.
In-vivo base editing of mutant huntingtin in a mouse model. Nature Biomedical Engineering (2026) (opens in a new tab). Dual-AAV9, direct-striatal preclinical study.
Nelson et al. In vivo adenine base editing ameliorates Dravet syndrome phenotypes in a mouse model. Science Translational Medicine (2026) (opens in a new tab).
ClinicalTrials.gov: ETX101 in SCN1A-positive Dravet syndrome (NCT05419492) (opens in a new tab). AAV9-delivered transcriptional regulation, not DNA editing.
APOE4-to-APOE3 prime-editing study in induced neurons and mouse models. Advanced Science (2026) (opens in a new tab). Preclinical evidence.
ClinicalTrials.gov: HG204/HERO study for MECP2 duplication syndrome (NCT06615206) (opens in a new tab). Intracerebroventricular Cas13 RNA targeting, not DNA editing.
Nonhuman-primate study of AAV central-nervous-system tropism. Nature Nanotechnology (2023) (opens in a new tab). Evidence for age-, species- and cell-type-dependent transduction.
National Institute of Mental Health: Brain Stimulation Therapies (opens in a new tab). Clinical uses, course and adverse-effect overview.
FDA 510(k) summary K210201: BrainsWay Deep TMS System (opens in a new tab). Anxiety-symptom reduction within treatment of adult major depressive disorder.
FDA 510(k) summary K231926: NeuroStar Advanced Therapy System (opens in a new tab). Adjunctive major-depressive-disorder indication for ages 15–21.
FDA 510(k) record K200957: BrainsWay Deep TMS System (opens in a new tab). Short-term smoking-cessation indication.
FDA 510(k) summary K260402: MeRT System (2026) (opens in a new tab). Adult adjunctive PTSD indication and evidentiary limitations.
FDA 510(k) summary K162797: SpringTMS Total Migraine System (opens in a new tab). Prescription single-pulse TMS for acute and preventive migraine treatment.
FDA PMA record P230024: Flow FL-100 (2025) (opens in a new tab). Approval date and major-depressive-disorder indication.
Rossi et al. Safety and recommendations for TMS use. Clinical Neurophysiology (2021) (opens in a new tab). Expert safety guidelines.
FDA: Flow FL-100 Summary of Safety and Effectiveness Data (opens in a new tab). Pivotal results, adverse events, benefit uncertainty and labeled protocol.
Antal et al. Updated safety and ethics guideline for low-intensity transcranial electrical stimulation (2026) (opens in a new tab). Supervised and remote-use safeguards.
Meta-analysis of offline high-frequency rTMS and cognition in healthy participants. Neuropsychology Review (2024) (opens in a new tab).
Umbrella review of prefrontal tDCS effects on cognition (2022) (opens in a new tab). Review of conflicting evidence.
Vigne et al. BDNF Val66Met and clinical response to rTMS in depression (2026) (opens in a new tab). Naturalistic cohort finding no response or remission association.
Lustenberger et al. Feedback-controlled transcranial alternating-current stimulation and sleep spindles. Current Biology (2016) (opens in a new tab). Small human closed-loop proof-of-concept.
Sahel et al. Partial recovery of visual function after optogenetic therapy. Nature Medicine (2021) (opens in a new tab). Human retinal optogenetics with viral gene delivery and goggles.
Barksdale et al. Low-intensity focused ultrasound targeting the human amygdala. Molecular Psychiatry (2025) (opens in a new tab). Target engagement and preliminary therapeutic investigation.
Flesher et al. A brain–computer interface that evokes tactile sensations improves robotic-arm control. Science (2021) (opens in a new tab).
UNESCO: Recommendation on the Ethics of Neurotechnology (2025) (opens in a new tab).
OECD: Recommendation on Responsible Innovation in Neurotechnology (opens in a new tab).
World Health Organization: Human genome editing (opens in a new tab). Governance resources and somatic, germline and heritable distinctions.
European Union: Implementing Regulation (EU) 2022/2347 (opens in a new tab). Risk classification for Annex XVI non-medical brain-stimulation equipment.
FDA: General Wellness—Policy for Low Risk Devices (opens in a new tab). Explains the agency’s policy boundary for low-risk general-wellness products.

Back to blog