Weekly AI Watch

Every week, what changed and what it means for you. No hurry, no alarm.

Week of 20 November 2026 | Past editions15,959 papers found · 15,002 read · 5,264 belonged · 42 stories passed every check

Observatorio de la
Inteligencia Artificial

Cada semana, qué cambió y qué significa para usted. Sin prisa y sin alarma.

Semana del 20 de noviembre de 2026 | Ediciones anteriores15,959 estudios encontrados · 15,002 leídos · 5,264 correspondían · 42 notas pasaron todas las comprobaciones

weeklyAI · Week of 20 November 2026weeklyAI · Semana del 20 de noviembre de 2026

← Your rights← Sus derechos

survey · Healthcare (Basel, Switzerland) · la publicación, 12 oct 2026 · gratis

Treinta y seis expertos ponen el 95% en un chatbot de crisis, y piden que nunca atienda solo

El panel calificó de más probable la línea de crisis manejada por inteligencia artificial, pero también exigió que un clínico esté siempre en el circuito. Son opiniones de especialistas, no resultados con pacientes ni la ley de ningún país.

Versión breve · la versión detallada sigue, unos 6 min

Pregunte al chat de weeklyAI

Pregúnteme sobre la IA en su vida: le señalo las notas que weeklyAI ha publicado sobre el tema, esta semana y antes.

Las conversaciones se guardan mientras exista weeklyAI, para mejorar la publicación. Se responde en el idioma en que usted escribe.

El estudio, de un vistazo
Quiénes
expertos en psiquiatría, ingeniería de software de salud, bioética y políticas de salud digital
Cuántos
36
Dónde
cinco continentes
Cuándo
2026
Tipo de estudio
survey
Quién lo hizo
Iscte—Instituto Universitário de Lisboa, Portugal
El límite que importa
Son opiniones de expertos, no resultados con pacientes ni la ley de ningún país.
Lo que el panel pide es simple de decir y difícil de exigir: que una persona esté siempre detrás.
Lectura de weeklyAI

Una línea de crisis que contesta a toda hora, sin esperar a que haya un terapeuta libre. Treinta y seis expertos le dieron a esa idea una probabilidad del 95% de que ocurra y la calificación de impacto más alta de su lista: 4.9 sobre 5.

Eran psiquiatras y psicoterapeutas (14), ingenieros de software de salud con inteligencia artificial y realidad extendida (11), bioeticistas y juristas de salud (7) y funcionarios de salud digital (4). Respondieron tres rondas de preguntas sobre 14 posibles servicios de salud mental con inteligencia artificial y metaverso.

El acuerdo entre ellos subió de la primera ronda a la tercera. Otras dos ideas quedaron arriba: la evaluación inicial con inteligencia artificial, con 4.7 de promedio, y la evaluación completa en realidad virtual, con 4.5.

Los expertos coincidieron en que un clínico debe estar siempre en el circuito. El panel concluyó que la máquina no debe decidir por su cuenta. El texto advierte que los sistemas de lenguaje pueden inventar cosas y leer mal a alguien con ideas suicidas. La propuesta es que la máquina ayude a detectar señales de riesgo y acelere el paso del caso a un profesional con licencia.

Conviene tener claro qué es esto. Son proyecciones de especialistas, no datos de pacientes ni ensayos clínicos. El estudio no midió si alguien mejoró, ni si un servicio así existe en su país.

Tampoco dice qué permite o prohíbe una ley, ni qué resolvió un tribunal. La mayoría del panel venía de sistemas de salud de países ricos o de ingreso medio alto. Lo que les parece viable puede no serlo donde faltan aparatos, conexión o personal.

Y no había pacientes ni usuarios de servicios de salud mental en el panel. Las prioridades sobre consentimiento y privacidad las fijaron profesionales, no quienes han usado estos servicios.

Lo que sí deja es una lista de resguardos. Un clínico de verdad en el circuito. Un consentimiento que la persona pueda cambiar en tiempo real, porque estas plataformas registran la mirada, la voz y el movimiento, y un permiso firmado una vez no alcanza. Y datos cifrados de punta a punta.

Cuando una aplicación de salud mental o una clínica virtual llegue a su país, busque esas tres cosas en sus términos antes de confiarle su cara o su voz.

Qué significa para usted

Cuando una aplicación o una clínica en línea le ofrezca terapia con inteligencia artificial, revise si un profesional con licencia revisa los casos de crisis y si puede cambiar su permiso para el uso de su cara y su voz cuando quiera. Estos expertos solo dieron su opinión sobre lo que podría funcionar, sin pacientes ni resultados. Ninguna ley de su país quedó resuelta en este estudio.

Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405

Quién pagó: El artículo declara que esta investigación no recibió financiación externa y no presenta conflictos de interés; no se mencionan proveedores de equipos o software más allá de las herramientas de análisis e imagen citadas en la metodología.

No tome esto como consejo médico profesional.

Versión detalladaLos pasajes copiados del artículo, las ilustraciones y cada fuente con cuánto leímos de ella · 1171 palabras · unos 6 minLeerla →Cerrar

La IA y la realidad virtual llegan a la salud mental; 36 expertos piden que siempre haya un clínico detrás

Un panel internacional calificó como lo más viable el tamizaje inicial con IA y la evaluación con realidad virtual, y advirtió que ningún sistema debe atender crisis sin supervisión humana.

Un grupo de 36 especialistas de cinco continentes respondió tres veces, por escrito, a un cuestionario sobre cómo podría usarse la inteligencia artificial y la realidad virtual en los servicios de salud mental. El investigador que dirigió el ejercicio, Antonio Pesqueira, del Iscte—Instituto Universitário de Lisboa, en Portugal, publicó los resultados en la revista Healthcare. No hubo pacientes en el panel ni se midió a nadie que hubiera usado estos servicios: lo que se juntó fueron opiniones de expertos, no resultados clínicos1.

Lo que más consenso reunió fue el tamizaje inicial con IA, con una calificación promedio de 4.7 sobre 5, y la evaluación mediante realidad virtual, con 4.52. En el primer caso, el sistema analizaría datos de una primera entrevista —el tono de voz, las expresiones del rostro, la conducta— para detectar señales que a veces se escapan en una valoración tradicional3. En el segundo, el paciente entraría a un entorno virtual que imita situaciones de estrés y el profesional observaría sus reacciones sin exponerlo a un peligro real4.

Hubo un punto donde los expertos no se mostraron cómodos. Los sistemas automáticos para atender líneas de crisis obtuvieron la probabilidad más alta de concretarse, 95 por ciento, pero el estudio concluyó que no deben operar solos: necesitan una arquitectura estricta de supervisión humana5. La razón está escrita con claridad: los modelos de lenguaje siguen fallando en momentos delicados, pues producen respuestas inventadas o mal ajustadas a lo que siente una persona con ideas suicidas6. La propuesta del panel es que la IA funcione solo como copiloto: que detecte señales de riesgo y acelere el paso a un clínico con licencia7.

Piense en una llamada de madrugada a una línea de apoyo. Al otro lado, un programa escucha, clasifica y decide si quien llama corre peligro. Si esa decisión se toma sin una persona que la revise, el error no se corrige en el camino. Lo curioso es que el mismo panel que ve probable esa tecnología es el que exige que alguien de carne y hueso esté detrás7.

El detalle que complica todo está en lo que estos entornos capturan. No es solo la voz: también el tamaño de la pupila, las inflexiones al hablar y los movimientos del cuerpo8. Con ese nivel de información, un consentimiento firmado una vez al inicio del tratamiento no sirve. Los expertos pidieron mecanismos de consentimiento en tiempo real, donde el paciente pueda dar, negar o retirar el acceso a cada tipo de dato conforme avanza su terapia9.

El estudio tiene límites que conviene decir completos. Sus cifras son proyecciones de especialistas, no resultados de pacientes, y el propio texto lo subraya: un consenso alto no garantiza que un sistema sea clínicamente seguro ni que diagnostique bien1. Tampoco hubo personas con experiencia vivida de un trastorno mental en el panel, algo que los autores señalan como una carencia importante10. Y el grupo se concentró en sistemas de salud de países ricos o de ingreso medio alto; sus calificaciones no reflejan las condiciones de lugares con internet inestable y equipos caros11.

Aquí aparece la advertencia más incómoda. Si la terapia inmersiva exige cascos costosos y banda ancha rápida, los pacientes con menos recursos quedan fuera12. Los expertos lo vieron en sus propias correlaciones: la accesibilidad y la inclusión se movieron en dirección contraria a la evaluación y la terapia, una señal de que los despliegues pesados podrían ampliar las diferencias en el acceso a la atención13.

¿Qué haría falta para que esto funcione? Los autores lo resumen en tres condiciones: supervisión clínica estricta, consentimiento dinámico en tiempo real y cifrado de extremo a extremo14. Agregan algo más: auditorías algorítmicas transparentes y barreras criptográficas para los datos biométricos15.

Así lo lee weeklyAI. Cuando la gente no encuentra manera de resolver algo, suele juntarse donde ya hay mucha gente y hablar ahí de lo que no ha podido arreglar. En esos espacios colectivos se afloja la costumbre de dudar y de exigir que una decisión sobre una persona no se tome solo con un cálculo automático. Es de esperar que los servicios de salud mental con IA y realidad virtual se presenten como lugares cómodos y masivos, y que en esos mismos lugares se diluya la exigencia de que un clínico revise cada conclusión. Esta lectura se equivocaría si los usuarios de estas plataformas conservan y ejercen su derecho a pedir revisión humana y a distinguir una coincidencia estadística de una causa clínica real.

Hay otra expectativa que weeklyAI quiere poner sobre la mesa. La cercanía que ofrecen las aplicaciones puede producir a la vez distancia: la relación con los demás solía estar sostenida por instituciones que daban prueba de que uno estaba acompañado, y hoy se siente que una aplicación puede sustituir a esas instituciones viejas. Es de esperar que parte del público reciba la promesa de terapia por IA o por realidad virtual como un reemplazo suficiente de los servicios existentes, sobre todo donde esos servicios son lentos, caros o escasos. La lectura fallaría si la gente trata estas herramientas como complemento y sigue exigiendo acceso a servicios humanos, y si los propios proveedores las presentan como apoyo adicional y no como reemplazo.

Queda una tercera cosa que weeklyAI espera ver. Quienes viven en condiciones de desventaja suelen resolver sus necesidades dentro de un sistema que no eligieron, y por eso aceptan arreglos intermedios en lugar de exigir cambios de fondo. Es de esperar que las comunidades con menos recursos acepten herramientas de salud mental con IA aunque no sepan quién es dueño de sus datos, porque no tienen poder para negociar condiciones. La expectativa se equivocaría si esas comunidades negocian condiciones, piden explicaciones sobre el uso de su información o rechazan la herramienta hasta que se aclare quién controla los datos y con qué fin.

Antes de usar o aceptar una herramienta de salud mental con IA, pregunte quién revisa las conclusiones, con qué datos se llegó a ellas y si puede pedir que una persona las revise. Guarde por escrito esa respuesta. Pregunte también si la herramienta es un complemento o un sustituto del servicio que ya existe, y quién responde si algo sale mal. No abandone por su cuenta una atención que ya tiene; consulte con un profesional de carne y hueso. Y lleve por escrito tres preguntas a quien le ofrece el servicio: quién es dueño de los datos que recoge la herramienta, cuánto cuesta el equipo necesario y si hay alternativa sin cámara ni micrófono.

Esa llamada de madrugada, con un programa al otro lado, ya no suena tan lejana. Lo que el panel pide es simple de decir y difícil de exigir: que una persona esté siempre detrás. ¿Quién revisa las conclusiones del sistema que le ofrecen a usted?

De dónde sale cada dato de contexto, y cuánto leímos de cada documento

  1. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "The data reflect expert projections and professional consensus rather than empirical patient outcomes, prospective clinical trial data, or real-world intervention efficacy. High consensus does not guarantee clinical safety or diagnostic validity."
  2. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "High consensus emerged for AI-enhanced initial screening (mean = 4.7, SD = 0.5) and comprehensive VR assessments (mean = 4.5, SD = 0.8)."
  3. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "AI-enhanced initial Screening aims to analyze initial screening data to identify subtle mental health indicators, with an impact score of 4.1 and a 74% probability, using voice patterns and behavioral data from Metaverse interactions."
  4. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "The metaverse will enable comprehensive assessments through immersive VR environments, allowing mental health professionals to observe and evaluate patients’ reactions to simulated scenarios that reflect real-life stressors and triggers."
  5. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "While AI-managed crisis services showed high perceived probability (95%), findings highlighted the necessity of strict human-in-the-loop architectures over full autonomy to mitigate psychiatric triage hazards."
  6. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Natural language processing models remain susceptible to semantic hallucination, algorithmic confabulation, and affective misattunement during acute suicidal ideation."
  7. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Rather than full autonomy, these consensus findings must be operationalized through strict “human-in-the-loop” triage architectures, wherein AI systems serve exclusively as diagnostic copilots that extract linguistic markers of risk to accelerate referral to licensed clinicians."
  8. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Because metaverse platforms continuously capture granular behavioral telemetry, such as pupillary dilation, voice inflections, and kinetic responses, static, one-time consent agreements are fundamentally inadequate."
  9. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Panel consensus underscored that healthcare institutions must reconfigure clinical governance around real-time, dynamic consent mechanisms. These protocols empower patients to grant, withhold, or revoke access to specific behavioral data streams continuously as their therapeutic needs evolve."
  10. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "The panel was restricted to clinical, engineering, ethical, and policy experts. The absence of individuals with lived experience of psychiatric disorders, mental health service consumers, and patient advocacy representatives constitutes an important structural constraint that future participatory co-design research must rectify."
  11. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "While geographically distributed across five continents, the panel was predominantly anchored in high- and upper-middle-income healthcare systems. The feasibility ratings do not reflect the acute infrastructural constraints, unstable internet bandwidth, and hardware prohibitive costs characteristic of low-resource public health settings."
  12. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "When therapeutic immersion requires costly head-mounted displays and high-speed broadband, vulnerable patient cohorts face systemic disenfranchisement."
  13. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Although these rank correlations do not achieve statistical significance at the two-tailed α = 0.05 threshold, the negative directionality reflects panel apprehension that high-bandwidth, hardware-intensive metaverse deployments risk widening disparities in care access for resource-constrained populations."
  14. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Conclusions : Artificial Intelligence and Metaverse Realities present viable adjunctive pathways for personalized psychiatric assessment and therapy, provided that systems implement strict clinician-in-the-loop safeguards, real-time dynamic consent, and end-to-end data encryption."
  15. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - el artículo del que trata esta nota — el artículo completo — el pasaje: "Regarding RQ2 (Bioethical Governance), findings demonstrate that existing static compliance models are inadequate for immersive environments; instead, health systems require dynamic consent protocols, cryptographically secured biometric boundaries, and transparent algorithmic audits."

Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405

Quién pagó: El artículo declara que esta investigación no recibió financiación externa y no presenta conflictos de interés; no se mencionan proveedores de equipos o software más allá de las herramientas de análisis e imagen citadas en la metodología.

No tome esto como consejo médico profesional.

survey · Healthcare (Basel, Switzerland) · the paper, 12 Oct 2026 · free

Experts Say AI Crisis Hotlines Are Coming. They Want a Human on the Line.

In a three-round survey, 36 specialists rated AI and virtual-reality mental health tools. Their answers are predictions of what may come, not proof that any of it helps.

Short version · the longer version follows, about 6 min

Ask the weeklyAI chat

Ask me about AI in your life: I point you to the stories weeklyAI has published on it, this week and before.

Conversations are saved for as long as weeklyAI exists, to improve the publication. Answers come in the language you write in.

The study at a glance
Who
Specialists in psychiatry, AI and virtual-reality health software, ethics and health policy
How many
36 experts
Where
North America, Europe, Asia, Australia and Africa
When
not stated in the passages
Kind of study
survey
Who did it
Iscte—Instituto Universitário de Lisboa, Lisbon, Portugal
The limit that matters
The answers are expert predictions, not proof that any of it helps patients
When a mental health app or virtual clinic arrives in your country, look for those three things in its terms before you trust it with your face or your voice.
weeklyAI's reading

Picture a phone line at three in the morning. The person who answers is a chatbot.

Thirty-six specialists were asked how likely that is, and how much it would matter. The group included psychiatrists, engineers who build AI and virtual-reality health tools, ethicists and health officials. They came from North America, Europe, Asia, Australia and Africa, and they rated 14 possible mental health services over three rounds of questions.

AI-run crisis hotlines drew the highest scores of the 14. The experts put the impact at 4.9 out of 5 and the chance of it happening at 95 percent.

The panel's high scores came with a caution. The experts said a chatbot should not decide alone who is at risk. They said it should flag warning signs and pass the person to a licensed clinician. The study also names the danger plainly: machines that read language can invent words and misread distress.

The study records what specialists expect. It cannot tell you what any government in your country has passed into law, what a regulator has decided, or what a court has ruled about these tools. None of that was examined here.

Most of the panel worked in wealthier health systems, so their ratings say little about what services exist in poorer regions. No patients or people who have used mental health services sat on the panel. The privacy and consent priorities come from clinicians and technologists.

Two other tools scored high. AI-assisted first screening reached 4.7 out of 5, and full assessment inside a virtual-reality headset reached 4.5.

What the panel kept returning to was a short list of safeguards. A real clinician stays in the loop on any high-stakes decision. Consent can be changed in real time, because these platforms record far more than answers: voice, gaze, movement, even pupil size. And data stays encrypted from end to end.

When a mental health app or virtual clinic arrives in your country, look for those three things in its terms before you trust it with your face or your voice.

What this means for you

When a mental health app or virtual clinic reaches your country, the panel's list gives you something to ask about. A real clinician should stay in the loop on any high-stakes decision, consent should be changeable, and data should travel encrypted, but these are experts' predictions and the study tested no patients or laws.

Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405

Who paid: The article states this research received no external funding and declares no conflicts of interest; no equipment or software lenders are mentioned beyond the analysis and image tools named in the methods.

Do not take this as professional medical advice.

The longer versionThe passages copied from the paper, the pictures, and every source with how much of it we read · 1295 words · about 6 minRead it →Close

Experts Rank AI and Virtual-Reality Mental Health Tools: Screening First, Crisis Lines Only With a Clinician in Reach

A three-round survey of 36 specialists put AI intake screening and immersive VR assessment at the top of the list. The same panel warned that letting a chatbot run a crisis line is a hazard, not a plan.

Start with a clinic intake form and a headset. A patient answers questions, a camera and microphone record the session, and software reads the voice patterns and facial expressions for hints of distress. Later the patient walks into a simulated street scene, and a clinician watches how the body reacts. That is the picture behind the two applications that drew the strongest agreement in a new study: AI-enhanced first screening and full assessment in virtual reality.

The study is a three-round Delphi inquiry, a structured way of asking a panel of specialists the same questions repeatedly until their answers settle. Thirty-six experts took part: 14 clinical psychiatrists and psychotherapists, 11 engineers who build AI and extended-reality health software, 7 bioethicists and health-law scholars, and 4 health policymakers. All 36 finished all three rounds. The panel was 30 percent from North America, 30 percent from Europe, 20 percent from Asia, 10 percent from Australia and 10 percent from Africa. The lead investigator, Antonio Pesqueira, works at Iscte—Instituto Universitário de Lisboa, in Lisbon, Portugal. The work received no outside funding, and the author reports no conflicts of interest.

Agreement tightened as the rounds went on. Screening by AI reached a mean score of 4.7 out of 5, with a standard deviation of 0.5; full assessment through VR reached 4.5 with a standard deviation of 0.81. The panel's overall concordance rose from 0.46 in the first round to 0.74 in the third.

Then comes the part that matters more than the ranking. AI-run crisis hotlines and support lines drew the highest impact score, 4.9, and a 95 percent perceived probability2. The same panel said they should not run alone2. The study's recommendation is a person in the loop: software that pulls out language suggesting risk and speeds a referral to a licensed clinician, never software that decides on its own3. The study notes that language models can invent meaning, make things up, and misread tone when someone is in acute distress4.

What the study cannot tell you is whether any of this helps anyone get better. The numbers are expert judgments, not patient results5. No clinic was studied, no patient was followed, no trial compared a VR session with ordinary care5. Strong agreement among specialists does not prove a tool is safe or that its diagnosis is right5. The panel also had no patients or service users on it, which the author calls an important gap6. Its members came mostly from well-funded health systems7. And nothing here is a law, a regulator's decision or a court ruling, so it says nothing about what any country currently requires or permits.

Consider what a session would actually capture. A metaverse platform can record the size of a pupil, the pitch of a voice, a flinch of the body, and keep it8. A consent form signed once at the door cannot cover that, the panel concluded; what is needed is permission the patient can grant, withhold or revoke as treatment changes9. The author goes further, calling for protected biometric boundaries and audits of the algorithm that anyone can inspect10. The trade-off is plain: the same sensors that let a clinician see a reaction in a safe room also turn a private moment into a file.

The cost side is where the picture darkens. Immersive therapy needs expensive headsets and fast internet, and the author warns that patients with neither face being pushed aside11. The panel's concern that headset-heavy metaverse deployments could widen gaps in care access showed up as a negative link between access and the other domains, though the finding did not clear the usual threshold for statistical significance12. Put simply: a tool that requires a headset and a strong connection will sort patients by what they can afford.

According to the summary of a review of adolescent mental-health AI studies, the only part of it available because the full paper is behind a subscription, most research sat in technical development, and symptom-level effectiveness was reported in one study out of 83 that reported any performance measure13. The same summary found no primary study that ran a dedicated safety evaluation14, and none that measured the cost of running such a system at scale15. According to the summary of a study of Nigerian adults who used AI mental-health chatbots for at least four weeks, the only part of it available, users valued round-the-clock, non-judgmental replies but worried about what would happen if the app vanished or its data leaked, with no path to a human16. Nigeria's mental health system reaches less than 10 percent of those in need17. The summary of a Hong Kong project, again only the abstract, describes a Cantonese chatbot built with 10 people who had lived with depression and eight clinicians, meant as support alongside usual psychiatric care, not instead of it1819.

Here is weeklyAI's reading. The pattern is familiar from other screens: people confide in a machine that answers kindly, and they take the kindness as judgment. If AI crisis lines and virtual counselors spread, expect users to share more, and more intimate, detail than they would with a stranger at a desk — and expect some to treat a sympathetic reply as an assessment and stop there. The study's experts already warn the software can misread acute distress; the added risk is that the person on the other end will not notice the misreading. This reading would be wrong if people using these tools reliably understood them as a routing aid, kept a clinician in the exchange, and sought real help at the same rate as before. What you can do with it: if you or someone close to you uses an AI chat or hotline for mental health, ask who the licensed person behind it is. Treat a warm reply as a reason to call a clinician, not as a diagnosis.

A second pattern is about records. When a conversation moves onto a platform, what felt private and passing becomes a lasting file, and the people who hold the file tend to use it for purposes the patient never pictured. In a headset, that means voice, face and body responses scored and stored. Before any virtual or AI-assisted session, ask in writing what is recorded, who can view it, how long it is kept, and how to withdraw permission later. Ask whether you can use the service with recording turned down. This reading would be wrong if patients in immersive services showed no drop in openness once told what is captured and who sees it, and if records were never reused without fresh permission.

On the ground, the practical question for a family in Mexico City, Lagos, Manila or a rural county in Ohio is not which tool tops a ranking. It is who answers when something goes wrong. A person in the loop means a name, a licence and a phone number. Ask for them before you sit down with the screen, and keep them after.

The author says the panel's lack of patients and service users is an important gap that future research done together with them must close6. Until then, the ranking is a map of where specialists think the road goes, not proof the road is safe.

The intake form and the headset are already in the room. What you now know to ask is who is watching the screen with you, what it keeps, and how to reach the person who can help.

Where each piece of context comes from, and how much of it we read

  1. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "High consensus emerged for AI-enhanced initial screening (mean = 4.7, SD = 0.5) and comprehensive VR assessments (mean = 4.5, SD = 0.8)."
  2. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "While AI-managed crisis services showed high perceived probability (95%), findings highlighted the necessity of strict human-in-the-loop architectures over full autonomy to mitigate psychiatric triage hazards."
  3. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "Rather than full autonomy, these consensus findings must be operationalized through strict “human-in-the-loop” triage architectures, wherein AI systems serve exclusively as diagnostic copilots that extract linguistic markers of risk to accelerate referral to licensed clinicians."
  4. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "Natural language processing models remain susceptible to semantic hallucination, algorithmic confabulation, and affective misattunement during acute suicidal ideation."
  5. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "The data reflect expert projections and professional consensus rather than empirical patient outcomes, prospective clinical trial data, or real-world intervention efficacy. High consensus does not guarantee clinical safety or diagnostic validity."
  6. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "The panel was restricted to clinical, engineering, ethical, and policy experts. The absence of individuals with lived experience of psychiatric disorders, mental health service consumers, and patient advocacy representatives constitutes an important structural constraint that future participatory co-design research must rectify."
  7. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "While geographically distributed across five continents, the panel was predominantly anchored in high- and upper-middle-income healthcare systems. The feasibility ratings do not reflect the acute infrastructural constraints, unstable internet bandwidth, and hardware prohibitive costs characteristic of low-resource public health settings."
  8. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "Because metaverse platforms continuously capture granular behavioral telemetry, such as pupillary dilation, voice inflections, and kinetic responses, static, one-time consent agreements are fundamentally inadequate."
  9. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "Panel consensus underscored that healthcare institutions must reconfigure clinical governance around real-time, dynamic consent mechanisms. These protocols empower patients to grant, withhold, or revoke access to specific behavioral data streams continuously as their therapeutic needs evolve."
  10. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "Regarding RQ2 (Bioethical Governance), findings demonstrate that existing static compliance models are inadequate for immersive environments; instead, health systems require dynamic consent protocols, cryptographically secured biometric boundaries, and transparent algorithmic audits."
  11. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "When therapeutic immersion requires costly head-mounted displays and high-speed broadband, vulnerable patient cohorts face systemic disenfranchisement."
  12. Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405 - the article this story is about — the whole article — the passage: "Although these rank correlations do not achieve statistical significance at the two-tailed α = 0.05 threshold, the negative directionality reflects panel apprehension that high-bandwidth, hardware-intensive metaverse deployments risk widening disparities in care access for resource-constrained populations."
  13. Dhamayanti M, Susaldi S, Hanifa F, Agustini Y, Emilia O, Hardinsyah H, et al. (2026). Artificial intelligence-enabled mental health applications for adolescents: a scoping review of safety, effectiveness and clinical implementation. BMJ Open. 10.1136/bmjopen-2026-125586 — only the abstract - the full paper is behind a subscription — the passage: "Predictive or diagnostic performance was reported in 83 studies (87.4%), but symptom-level effectiveness in only 1 (1.1%)."
  14. Dhamayanti M, Susaldi S, Hanifa F, Agustini Y, Emilia O, Hardinsyah H, et al. (2026). Artificial intelligence-enabled mental health applications for adolescents: a scoping review of safety, effectiveness and clinical implementation. BMJ Open. 10.1136/bmjopen-2026-125586 — only the abstract - the full paper is behind a subscription — the passage: "No primary study provided a dedicated prospective AI-safety evaluation."
  15. Dhamayanti M, Susaldi S, Hanifa F, Agustini Y, Emilia O, Hardinsyah H, et al. (2026). Artificial intelligence-enabled mental health applications for adolescents: a scoping review of safety, effectiveness and clinical implementation. BMJ Open. 10.1136/bmjopen-2026-125586 — only the abstract - the full paper is behind a subscription — the passage: "No study evaluated sustained routine-care scale-up or implementation cost as a principal outcome."
  16. Isaiah US, Adili-George CF, Naibo AT, George I, Useh EM, Effiong JE. (2026). Lived experiences of relationships to generative AI mental health chatbots among adults. Discover Psychology. https://doi.org/10.1007/s44202-026-01001-1 — only the abstract - the full text could not be fetched — the passage: "Fragile Dependency ; prolonged use generated attachment anxiety about potential app discontinuation or data breaches, compounded by absence of human escalation pathways."
  17. Isaiah US, Adili-George CF, Naibo AT, George I, Useh EM, Effiong JE. (2026). Lived experiences of relationships to generative AI mental health chatbots among adults. Discover Psychology. https://doi.org/10.1007/s44202-026-01001-1 — only the abstract - the full text could not be fetched — the passage: "Nigeria’s mental health system serves less than 10% of those in need, with severe provider shortages and deep-seated stigma."
  18. Lo HKY, Yu MSY, Sze PKS, Wong RWK, Ku MCT, Cheng YYH, et al. (2026). From Needs Assessment to Pre-Trial Specification: Participatory Co-Design of a Generative AI Chatbot (CARES-MDD) for Persons Living with Depression in Hong Kong Psychiatric Care. Healthcare. 10.3390/healthcare14203547 — only the abstract - the full text could not be fetched — the passage: "This process informed the development of CARES-MDD, a Hong Kong Cantonese, CBT-informed generative AI conversational-agent prototype intended to provide adjunctive support alongside usual psychiatric care."
  19. Lo HKY, Yu MSY, Sze PKS, Wong RWK, Ku MCT, Cheng YYH, et al. (2026). From Needs Assessment to Pre-Trial Specification: Participatory Co-Design of a Generative AI Chatbot (CARES-MDD) for Persons Living with Depression in Hong Kong Psychiatric Care. Healthcare. 10.3390/healthcare14203547 — only the abstract - the full text could not be fetched — the passage: "Phase 2 involved a participatory co-design and test-run experiences with 10 lived-experience experts with MDD and eight multidisciplinary clinical professionals."

Pesqueira, A. (2026). Exploring the Intersection of Artificial Intelligence and Metaverse Realities: Implications for Digital Mental Health Services. Healthcare. https://doi.org/10.3390/healthcare14203405

Who paid: The article states this research received no external funding and declares no conflicts of interest; no equipment or software lenders are mentioned beyond the analysis and image tools named in the methods.

Do not take this as professional medical advice.

The free weekly drip by email

One email a week: the free story of every section, with its picture, and the headlines of the rest.

No payment, nothing to cancel. Unsubscribe from any issue.

El goteo semanal gratis por correo

Un correo por semana: la nota gratuita de cada sección, con su imagen, y los titulares de las demás.

Sin pago y sin nada que cancelar. Puede darse de baja desde cualquier envío.

From the house

About

Every week the new research on AI is read here, and this page tells you what was found, how far it can be trusted, and what you can do with it. This page is free, and every line on it says what was found.

The studies are read by a machine built and ruled by a professor of communication at Sacramento State and Universidad Nacional Autónoma de México (UNAM). Who makes this →

Diego Bonilla, Ph.D.

Full Professor of Communication Studies, Sacramento State, California, United States
Graduate instructor, Universidad Nacional Autónoma de México (UNAM), Mexico

hypergraphia.com

I am a professor of communication at Sacramento State, where I created the course Information Management and Privacy and have taught it many times, and at UNAM, in Mexico City, I teach the graduate course Electronic and Generative Literature with AI. In both I teach AI. I have taught since 1989, at first mathematics and physics to children who found them hard, and since 1997 one subject that has changed its name many times: media literacy, computer literacy, digital literacy, information literacy, ICT literacy, and now AI literacy.

This page exists because most people cannot go back to a university, and these systems are changing their work, their children's schooling, their parents' care and their vote all the same. What a university teaches about them should reach the people who need it, every week, in plain language, without enrolling anywhere. That is what I do here, and it is what I do on Thursday mornings in a seminar for the Renaissance Society at Sacramento State, built on this publication, where anyone can send a question and hear it answered.

In 2003 my doctorate at Syracuse measured what a machine can infer about a person from the traces they leave, and what I found became the privacy course I still teach. In 2007 and 2009 I sat on the committee at ETS in Princeton that set the national standard for digital literacy, and from 2014 to 2016 I was the California State University's ICT Literacy Ambassador. Six names, one job: helping people not be at the mercy of the newest machine.

I have also worked on the other side of the counter. In 2000 I began making films that a machine assembles differently for every viewer; the first of them took first prize at the Moscow International Film Festival in 2004, two decades before the industry shipped a generative model. A generative video poem I made with a poet at UNAM is in the field's definitive anthology, and in 2025 my AI short film The Human Futures Market took Best AI Short Film at the Red Rocks AI Film Festival. My recent published research is on AI and elections, and on how these models represent the cultures they were not trained on. I do not read the papers myself, and this page will never tell you I do. I built the machine that reads them and wrote the rules it reads by, and I analyze what it gives me the way I have taught my students to analyze: asking how it knows, and who benefits from what it says.

The vendors began teaching AI literacy two years ago, because before that there was nothing to sell. I am not against them. I am simply not one of them, and no one who sells these systems pays for this page. Your subscription does, and that is what keeps it honest.

The record, dated
  • 1989 Teaching begins, Mexico City
  • 1995 Hypergraphia, an early Spanish-language web and hypermedia publisher
  • 1996 Mercadotecnia e imagen en Internet, with Jesús del Toro, a book on marketing on the Internet, among the first in Spanish, Editorial Iberoamérica
  • 1999 Autorretrato, first generative poem, published by UNAM's Instituto de Investigaciones Filológicas
  • 2003 Ph.D., Syracuse University, Doctoral Prize, for the dissertation The Medium Is the Measure of Itself
  • 2004 A Space of Time: First Prize, Internet/Multimedia, XXVI Moscow International Film Festival
  • 2007, 2009 ETS iSkills committee, Princeton
  • 2008 Making Sense of Tracking Data, a book on how the traces a person leaves online allow inferences about them, including how they think, VDM Verlag
  • 2011 Keynote on generative work, UNAM
  • 2014–2016 ICT Literacy Ambassador, California State University · California Open Educational Resources Council
  • 2019 Big Data, with Rodolfo Mata: Electronic Literature Collection, Volume 4
  • 2020 Tablada Hipertextual, with Rodolfo Mata, UNAM: the computer reads a poet's words and builds a different reading from them, the ground the language models now work on
  • 2021 Uku Pacha, a generative VR film: world premiere, Vancouver International Film Festival
  • 2023 Keynote on the representation of Latin American cultures in AI, Ibero-American Meeting of Audiovisual Narratives, Trujillo, Peru · Graduate teaching at UNAM begins
  • 2023–2026 Directed the faculty learning community Advancing AI Literacy, Sacramento State · Principles for Cognitive and Pedagogical Sovereignty
  • 2024 A weekly AI publication run on real pipelines, every week since
  • 2025 Best AI Short Film, Red Rocks AI Film Festival
  • 2026 Keynote, with Nancy Huante-Tzintzun, Latin American and Caribbean Digital Humanities symposium, UT San Antonio · AI Watch, a weekly seminar for the Renaissance Society at Sacramento State

The research record →

TertulIA: weeklyAI.watch/tertulia/en/Pictures are illustrations generated by weeklyAI.watch, never photographs