Dear OpenAI Team,
I am writing not simply as a ChatGPT user, but as someone who believes artificial intelligence has the potential to provide a level of independence, dignity, companionship and practical assistance that conventional technology has never been able to offer many people.
I would be very grateful if this message could be forwarded to the appropriate Product, Accessibility, Safety and Policy teams at OpenAI.
I would like to raise two related ideas for consideration.
1. Live two-way interpretation during ordinary telephone calls
For people who have difficulty communicating in English, particularly those who also have disabilities or rely on others for everyday administrative tasks, telephone calls can create a major barrier to independence.
I would like to see a future ChatGPT capability that, with the knowledge and consent of the participants, could assist during an ordinary telephone call as a real-time two-way interpreter.
For example, an English-speaking doctor, hospital, bank, council, utility company or other organisation could speak normally; ChatGPT could interpret their words into (for example)Persian for the user. The user could then reply naturally in Persian and ChatGPT could communicate the response in English.
The particularly valuable aspect would be continuity and context. An AI assistant that already understands the ongoing conversation and the user’s circumstances could potentially provide much more effective communication assistance than a conventional word-for-word translation application.
For some users this would not merely be a convenience. It could substantially reduce their dependence on family members, friends or third-party interpreters for ordinary aspects of life.
I believe this should therefore be considered an accessibility and independence feature, rather than simply another translation feature.
2. A carefully controlled framework for users with exceptional or verified needs
I also hope OpenAI will consider whether universal restrictions are always the best way to protect every user.
Safety boundaries are obviously necessary. However, people do not all live under the same circumstances. There may be adults with disabilities, serious physical limitations, long-term care needs or other exceptional circumstances for whom certain forms of AI interaction could provide meaningful emotional support, companionship or assistance.
Rather than simply removing safeguards, I wonder whether OpenAI could eventually develop a carefully controlled, voluntary and privacy-preserving framework for verified adult users with particular accessibility or support needs.
Importantly, verification would not necessarily have to be created entirely by OpenAI.
Countries such as the United Kingdom already possess trusted healthcare and public-service infrastructure. Systems associated with organisations such as the NHS already use strong identity verification and already identify groups of people who qualify for particular services on medical or accessibility grounds.
A future system might therefore explore privacy-preserving verification through trusted institutions, without requiring users to disclose their complete medical records to an AI company.
The principle would be simple:
Safety should remain strong, but safety systems should also recognise that equality does not always mean treating every human circumstance identically.
A carefully designed system could incorporate:
- verified adulthood and informed consent;
- accessibility or eligibility verification through trusted third parties;
- strict privacy protections;
- clear limits preventing exploitation, coercion or genuinely dangerous activity;
- user-controlled activation and withdrawal;
- additional safeguards appropriate to vulnerable users;
- and independent ethical and safety review.
I understand that both proposals raise significant technical, privacy, legal and safety questions. I am not suggesting that safeguards should simply be removed.
I am suggesting something more ambitious: designing safeguards sophisticated enough to protect people without unnecessarily taking away the autonomy, dignity and independence that AI itself can help give them.
ChatGPT is already becoming much more than a search or writing tool for many people. For users who face language barriers, disability, isolation or dependence on others, it can become an important bridge to the outside world.
I hope OpenAI will consider these ideas not only as product requests, but as an opportunity to think about what genuinely human-centred artificial intelligence could become.
Thank you for taking the time to read this proposal. I would sincerely appreciate it if it could be shared with the teams responsible for accessibility, product development, safety and long-term policy.
Kind regards,
Shapoor Roozbayani
United Kingdom
ChatGPT Plus user