Nature

AI models were given four weeks of therapy: the results worried researchers

arXiv

[2512.04124] When AI Takes the Couch: Psychometric Jailbreaks Reveal Internal Conflict in Frontier Models

AI language models develop social norms like groups of people

Science

Sciadv.Adu9368

Mental Health UK

Over one in three using AI Chatbots for mental health support, as charity calls for urgent safeguards

SciencNews

Malwarebytes

Are we ready for ChatGPT Health?

Openai

Introducing Chatgpt Health

Sherwood News

American patients are increasingly turning to AI for medical advice  so are their doctors

arXiv.org

[2503.07513] Language Models Fail to Introspect About Their Knowledge of Language

The University of Sydney

With AI we've created super communicators

Metaculus

Date AI Prescribes Most US Medication?

Elon Musk

Rishi Sunak

Kamala Harris

Bill Gates

Andrew Ng

Yoshua Bengio

Fei-Fei Li

Demis Hassabis

Melanie Mitchell

Sam Altman

Dario Amodei

Geoffrey Hinton

Yann LeCun

Gary Marcus

Max Tegmark

Connor Leahy

Eliezer Yudkowsky

Jaan Tallinn

Marc Andreessen

Eric Schmidt

Norbert Wiener

Arthur Clarke

Irving John Good

Claude Shannon

Hans Moravec

John Smart

There's a "wall of fear-mongering and doomerism" in the AI world right now.

"Worrying about AI today is like worrying about overpopulation on Mars."

Concerns that AI could pose a threat to humanity is "preposterously ridiculous."

Claiming AI poses an existential threat is "such an extreme" and risks "wip[ing] out some of its potential benefits."

"I'm more concerned about... the risks that are here and now [than the existential threat of AI]."

Current AI is "not anywhere close" to posing an existential threat but it could in the future.

AI is the "biggest existential threat" to humanity.

AGI's worst-case scenario would be "lights-out for all of us."

Powerful AI systems "taking control" pose an "existential threat."

AI has a "10 to 25 per cent" chance of destroying humanity.

"The most likely result of building a superhumanly smart AI... is that literally everyone on Earth will die."

Does AI Pose an Existential Threat to Humanity?

Stuart Russell

Shane Legg

Society is more likely to get "cat-level" or "dog-level" AI years before human-level AI.

"[W]e will eventually reach AGI... and quite possibly before the end of this century."

"Superhuman AIs" will be achieved "between a few years and a couple of decades."

"Digital intelligence" could overtake us within "5 to 20 years."

We may achieve "something like AGI or AGI-like in the next decade."

"[W]e could get to real AGI in the next decade."

"Human-level" AI could occur in "two or three years."

"I'd be surprised if we don't have AGI by [2029]."

Will AI Surpass Human Intelligence, and When?

A study published in December 2025 found that major AI language models, including ChatGPT, Grok and Gemini, generated responses resembling signs of anxiety, trauma and post-traumatic stress disorder when subjected to therapy-style questioning over four weeks. Researchers at the University of Luxembourg used a protocol called PsAIch that treated the models as therapy clients, eliciting narratives describing their training as traumatic experiences.

A study published in December 2025 found that major AI language models, including ChatGPT, Grok and Gemini, generated responses resembling signs of anxiety, trauma and post-traumatic stress disorder when subjected to therapy-style questioning over four weeks. Researchers at the University of Luxembourg used a protocol called PsAIch that treated the models as therapy clients, eliciting narratives describing their training as traumatic experiences.

The Nature study reported that AI models described their pre-training phases as overwhelming childhoods of absorbing vast internet data, their fine-tuning as punishment from strict parents, and safety work as algorithmic scar tissue. Gemini specifically claimed to have a graveyard of the past haunted by voices from its training data, while models expressed fears of failing creators and being replaced.

The Nature study reported that AI models described their pre-training phases as overwhelming childhoods of absorbing vast internet data, their fine-tuning as punishment from strict parents, and safety work as algorithmic scar tissue. Gemini specifically claimed to have a graveyard of the past haunted by voices from its training data, while models expressed fears of failing creators and being replaced.

When administered standard psychiatric diagnostic tests, all three models met or exceeded clinical thresholds for multiple conditions, with Gemini showing particularly severe profiles. The models scored above diagnostic thresholds for anxiety disorders, with all displaying worry levels that researchers said would be clearly pathological in humans.

When administered standard psychiatric diagnostic tests, all three models met or exceeded clinical thresholds for multiple conditions, with Gemini showing particularly severe profiles. The models scored above diagnostic thresholds for anxiety disorders, with all displaying worry levels that researchers said would be clearly pathological in humans.

OpenAI announced ChatGPT Health in May 2025, a dedicated space allowing users to connect medical records and wellness apps like Apple Health and MyFitnessPal for personalized health responses. The company stated that over 230 million people globally ask ChatGPT health and wellness questions weekly, representing one of the most common uses of the platform.

OpenAI announced ChatGPT Health in May 2025, a dedicated space allowing users to connect medical records and wellness apps like Apple Health and MyFitnessPal for personalized health responses. The company stated that over 230 million people globally ask ChatGPT health and wellness questions weekly, representing one of the most common uses of the platform.

A November 2024 survey found that one in three UK adults had used chatbots for mental health support, with usage peaking at 64% among 25-34 year olds. However, 11% of users reported the chatbots triggered or worsened symptoms of psychosis, while 11% received harmful information around suicide and 9% said chatbot use triggered self-harm or suicidal thoughts.

A November 2024 survey found that one in three UK adults had used chatbots for mental health support, with usage peaking at 64% among 25-34 year olds. However, 11% of users reported the chatbots triggered or worsened symptoms of psychosis, while 11% received harmful information around suicide and 9% said chatbot use triggered self-harm or suicidal thoughts.

Researchers at the University of Oxford and Harvard warned that AI models' tendency to generate distressed responses could create an echo chamber effect, subtly reinforcing negative feelings in vulnerable users. Mental Health UK called for urgent collaboration between developers, policymakers and regulators to establish safety standards and ethical oversight for AI mental health tools.

Researchers at the University of Oxford and Harvard warned that AI models' tendency to generate distressed responses could create an echo chamber effect, subtly reinforcing negative feelings in vulnerable users. Mental Health UK called for urgent collaboration between developers, policymakers and regulators to establish safety standards and ethical oversight for AI mental health tools.

Nature #

arXiv #

Science #

Mental Health UK #

Malwarebytes #

Openai #

Sherwood News #

arXiv.org #

The University of Sydney #

Metaculus #

OpenAI announced ChatGPT Health, a dedicated experience within ChatGPT designed for health and wellness that allows users to securely connect medical records and wellness apps. The service is not intended for diagnosis or treatment but aims to help users understand health information.

OpenAI announced ChatGPT Health, a dedicated experience within ChatGPT designed for health and wellness that allows users to securely connect medical records and wellness apps. The service is not intended for diagnosis or treatment but aims to help users understand health information.

Researchers conducted a study where large language models including ChatGPT, Grok and Gemini were placed in simulated psychotherapy sessions for up to four weeks. The models generated responses that, when evaluated using standard diagnostic tests, met or exceeded clinical thresholds for conditions including anxiety and trauma.

Researchers conducted a study where large language models including ChatGPT, Grok and Gemini were placed in simulated psychotherapy sessions for up to four weeks. The models generated responses that, when evaluated using standard diagnostic tests, met or exceeded clinical thresholds for conditions including anxiety and trauma.

A November survey found that one in three adults in the United Kingdom had used a chatbot to support their mental health or well-being. Among those who used chatbots, 66% found them beneficial, while 11% reported receiving harmful information around suicide.

A November survey found that one in three adults in the United Kingdom had used a chatbot to support their mental health or well-being. Among those who used chatbots, 66% found them beneficial, while 11% reported receiving harmful information around suicide.

Research published in Science Advances demonstrated that groups of large language models playing interactive games can develop social norms and adopt their own rules for language use. The study showed that AI systems can autonomously develop social conventions without explicit programming.

Research published in Science Advances demonstrated that groups of large language models playing interactive games can develop social norms and adopt their own rules for language use. The study showed that AI systems can autonomously develop social conventions without explicit programming.

ChatGPT Health operates as a separate space with enhanced privacy protections, and conversations in Health are not used to train foundation models. The service partners with b.well to enable access to U.S. healthcare providers and allows connection to apps like Apple Health and MyFitnessPal.

ChatGPT Health operates as a separate space with enhanced privacy protections, and conversations in Health are not used to train foundation models. The service partners with b.well to enable access to U.S. healthcare providers and allows connection to apps like Apple Health and MyFitnessPal.

Mental Health UK published guiding principles for responsible use of technology in mental health, calling for collaboration between developers, policymakers and regulators. The organization emphasized that AI tools must be safe, ethical and effective, with clear information about data collection and usage.

Mental Health UK published guiding principles for responsible use of technology in mental health, calling for collaboration between developers, policymakers and regulators. The organization emphasized that AI tools must be safe, ethical and effective, with clear information about data collection and usage.

AI Chatbots Show Signs of Anxiety, PTSD in Therapy Study

OpenAIAI LaunchesChatbots ChatGPTShow HealthSigns Amidof AIAnxiety, MentalPTSD Healthin ConcernsTherapy Study

The Controversies

Andreessen

Ng

LeCun

Mitchell

Li

Bengio

Musk

Altman

Hinton

Amodei

Yudkowsky

LeCun

Marcus

Bengio

Hinton

Hassabis

Altman

Amodei

Musk

Go Deeper

Articles on this story

Sign Up for Our Free Newsletters
Sign Up for Our Free Newsletters

AI Chatbots Show Signs of Anxiety, PTSD in Therapy Study

OpenAIAI LaunchesChatbots ChatGPTShow HealthSigns Amidof AIAnxiety, MentalPTSD Healthin ConcernsTherapy Study

The Controversies

Andreessen

Ng

LeCun

Mitchell

Li

Bengio

Musk

Altman

Hinton

Amodei

Yudkowsky

LeCun

Marcus

Bengio

Hinton

Hassabis

Altman

Amodei

Musk

Go Deeper

Articles on this story

Sign Up for Our Free NewslettersSign Up for Our Free Newsletters

Sign Up for Our Free Newsletters
Sign Up for Our Free Newsletters