PRODUCT PROPOSAL TO OPENAI
L. EEF – Always Within Reach
An integrated AI feature for safety, accessibility and emotional support
Initiated by: Tirzah
Submitted to: OpenAI – ChatGPT Product Development Team
-–
1. Introduction – Technology in the Service of People
Dear OpenAI Team,
I would like to bring a product concept to your attention that I believe could make a meaningful difference in the lives of many people.
As a regular ChatGPT user, I have experienced how valuable an AI assistant can be. Not only for information and practical support, but also for conversation, structure, emotional support and finding solutions during difficult moments.
This experience inspired the idea for:
L. EEF – Always Within Reach.
The principle behind this concept is simple: technology should support people not only when they are able to help themselves, but especially when they are temporarily or permanently unable to do so.
I propose developing L. EEF as an additional, optional feature integrated into ChatGPT.
An essential principle of this concept is that technology should serve people, not the other way around.
We recognise that there is also fear, uncertainty and distrust surrounding artificial intelligence in society. Sometimes these feelings arise because people do not fully understand what AI does, what it can do, or where its limitations lie.
With L. EEF, we want to demonstrate how AI can be used in a positive, responsible and human-centred way.
This is not about surveillance, constant monitoring or taking control away from people. It is about providing support when someone needs it.
The user remains in control:
The feature can be switched on and off by the user.
Users decide which functions they wish to activate.
Personal information is protected.
Contact with other people takes place only with prior consent.
AI does not make unlimited independent decisions on behalf of the user.
Technology that is there when you need it, not the other way around.
That is the philosophy behind L. EEF.
2. A Feature the User Controls
L. EEF should be an optional feature that users can independently switch on and off.
All existing ChatGPT features would remain available as usual. The additional feature would provide enhanced capabilities in safety, accessibility and personal support.
Users would decide which components to activate and which contacts may be approached, subject to prior consent.
3. Emergency Mode – Voice-Activated Emergency Assistance
One of the key components is hands-free emergency assistance.
Imagine someone walking with a rollator who falls, or someone who finishes showering but is unable to get up independently from their shower chair.
Their phone is out of reach. They simply call out:
“Eef, help. I can’t get up. Call my daughter.”
L. EEF could then:
Recognise and respond to the request for help.
Ask calm, targeted questions.
Help assess the severity of the situation.
Call a previously designated emergency contact.
Initiate a secure procedure to contact emergency services (112 in Europe) in an acute emergency.
Share the user’s location and relevant information.
Remain in voice communication with the user until help arrives.
Reliable voice recognition is essential, including when someone is distressed or has difficulty speaking.
4. Multilingual Support
Access to emergency assistance should never depend on the language someone speaks.
Users speaking Farsi, Arabic, Polish, English, Dutch or other languages should be able to request assistance in their own language.
L. EEF should then be able to communicate this information clearly and accurately to emergency responders.
Language should never become an unnecessary barrier to receiving help.
5. Emotional Support – Support During Difficult Moments
Not every request for help is a medical emergency.
Many people experience anxiety, panic attacks, grief, loneliness or moments when they feel overwhelmed and lose their sense of direction.
L. EEF should also be available outside Emergency Mode as an emotional support feature.
Possible functions include:
Guidance during a panic attack.
Gentle breathing exercises and grounding techniques.
A non-judgemental listening experience.
Help organising thoughts and feelings.
Support during anxiety and emotional distress.
Assistance in contacting a trusted person.
The assistant should adapt to the user’s emotional state: calm, patient, understandable and without overwhelming them with too much information.
6. Guidance & Connection – From Insight to Action
A third essential component is the ability to help people take meaningful next steps.
Sometimes people recognise that they need help but struggle to take the first step independently.
L. EEF could assist with:
Recognising and discussing recurring difficulties.
Identifying appropriate next steps.
Preparing for a conversation with a GP or healthcare professional.
Arranging appointments.
Drafting and, with permission, sending a message to a trusted person.
Finding appropriate professional support.
When there are signs of serious psychological distress or immediate danger, the assistant should ask careful questions and help connect the user with appropriate human support.
AI should complement human care, never replace it.
7. Accessibility and Independence
This feature could be particularly valuable for, among others:
People with neurological conditions.
People with physical disabilities.
People with limited mobility.
People with temporary disabilities.
People with intellectual disabilities.
Older adults and people living alone.
People who struggle with the local language.
People who are unable to operate their phones independently during periods of stress or panic.
For people with intellectual disabilities, it is important that the assistant can communicate information in an accessible way, take individual needs into account and remain patient, calm and easy to understand.
The aim is to increase independence, accessibility and a sense of safety, regardless of an individual’s abilities or circumstances.
8. Safety, Privacy and Reliability
Clear safety and privacy safeguards are essential for a feature of this nature.
Mutual, Explicit Consent
Users must explicitly consent to activating L. EEF and its associated functions.
Every designated emergency contact must personally agree to being registered in this role.
Emergency contacts must understand what the role involves, when they may be contacted and what information may be shared.
Emergency contacts must be able to withdraw their consent at any time.
Users must be able to easily appoint an alternative contact.
Verification of Emergency Contacts
L. EEF should be able to verify that a registered telephone number is active and in use.
Registration must be confirmed by the designated contact.
Periodic confirmation should establish whether the contact is still willing and available to fulfil this role.
Users must be notified if a number is no longer reachable or a contact withdraws their consent.
Privacy and Data Protection
Personal, medical and location data must be carefully protected.
Users must know when the microphone is active.
There must be a clear distinction between ordinary conversations and an activated safety mode.
Data may only be used in accordance with the agreed arrangements and permissions granted by the user.
Safety and Reliability
There must be a clear on/off function.
Safeguards must prevent accidental activation and false alarms.
There must be a procedure for situations in which a user stops responding.
The system must account for empty batteries, connectivity issues and technical failures.
Users must receive clear warnings whenever L. EEF is unavailable or not functioning correctly.
Clear boundaries must define which actions the AI is permitted to perform independently.
Human Assistance and Responsibility
An emergency feature must undergo extensive testing before it can be used to contact emergency services.
L. EEF must never give users the impression that help is on its way unless this has actually been confirmed.
Existing emergency services and professional care must never be replaced.
9. A Specific Request to OpenAI
We would like to ask OpenAI to have this concept evaluated by its product development team.
Would it be possible to develop L. EEF as an integrated safety, accessibility and support feature within ChatGPT?
We would appreciate feedback on:
Whether this concept is technically and practically feasible.
Which existing capabilities could be used to implement it.
What additional technological developments would be required.
Whether there are opportunities to further develop or explore this idea as a future product concept.
10. Closing Statement
The purpose of this idea is not to create yet another AI that knows everything.
It is about creating an AI that can also be there when someone is temporarily unable to help themselves.
An assistant that does not merely provide answers, but helps people reach the support they need.
One that does not simply listen, but also helps people take action.
And one that can give people, regardless of their physical abilities, intellectual abilities or native language, a greater sense of independence and safety.
Technology should not only make life easier. It should also help people stay alive, and above all, continue to truly live.
Kind regards,
Tirzah
Initiator of the concept
L. EEF – Always Within Reach
Live. Because everyone deserves the opportunity to live, with support whenever it is needed.