AI and People With Brain Injury

Generative AI is AI that creates content, e.g., text, images, video, music, and so on. I, a human being with a brain injury, am writing this post. Psychology Today requires its bloggers to confirm they used no artificial intelligence (AI) before submitting posts for review. I endorse this statement:

“Psychology Today is the international authority on behavioral science in large part because of your unique insight and breadth of knowledge. Using generative AI abdicates this expertise and authority, devalues Psychology Today and impugns your own reputation as a writer and subject matter expert.”

But not all AI is generative, even though generative AI developers have disrupted and dominated the conversation, affecting us all and creating three issues that apply particularly to people with brain injury. According to Galileo AI, which observes and evaluates GenAI apps, agentic AI autonomously pursues goals through “planning, tool use, and execution.” Attie, an alternative to closed AI ecosystems, is an open-source agentic AI that uses Anthropic to generate Attie’s responses that connect to your Atmosphere (ATProto) identity for “exploring and building on the network.” This kind of AI could emancipate people with brain injury wholly or partially from depending on others for daily tasks and online social life. But allowing AI in swirls up questions around privacy and medical confidentiality.

When you have a brain injury with slow processing, poor working memory, and/or concentration deficits, you’ll struggle with understanding why open-source AI empowers you unlike closed ecosystems and how to use AI beneficially while safeguarding your personal information and health data. Organizations that serve people with brain injury may not have thought about AI or don’t understand it and thus cannot guide their clients in safe and effective use. Still, knowing how companies like Meta and Google data mine and how to protect against their intrusions serves as a model for retaining confidentiality when using AI.

Client Privacy

AI requires a large amount of computing power. Neural processing units aid in using AI on laptops, but for the average person, AI computing is done through the cloud, that is, on data servers. EFF’s Surveillance Self-Defense notes the risk: “Depending on the type of tool you’re using, this can mean the company behind that tool can collect a variety of information, ranging from every query you type into a chatbot to the full text of a document or the entirety of a transcribed meeting.” If a privacy statement uses the phrase “reduce the amount of personal information," this does not mean the company eliminates personal information. In the future, someone using the trained model could recover it. So using AI offline on a laptop better safeguards medical information from privacy-breaking initiatives or from the mining and selling of your data.

In that case, the data a person wants an agentic AI to use must also be stored on the laptop: calendar, email, reminders, writing software, documents, images, videos, social media posts, blog posts, etc.

But let’s say you use AI on your phone or old laptop while erroneously assuming the phone’s AI doesn’t depend on the cloud. As Surveillance Self-Defense notes: The only way to determine what is on the cloud is to “try disabling the internet connection on your device then using the AI feature. If it still works, it’s likely an on device feature.” A healthcare provider could then discuss with the client how to adapt their use.

Healthcare Professional-Client Interaction

Some clients are already using AI agents on their phones and laptops and may not consider their health team or caregiver’s privacy when using agentic AI in their homes. They may film them executing a task, ask AI to store and analyze the video, and later ask their AI to replay the video for them to follow. When agentic AI uses cloud computing, the caregiver’s face will be uploaded to the cloud without their permission, along with their name and any other information on them that their client has stored in their contacts or notes. Companies may tie this data to other online data, such as where a healthcare professional shops, their friends and family, which clients they visit, which clinics they work at, and so on. Some people with brain injury may not understand the need for their healthcare team’s or caregiver’s privacy and may swoop in with their phone and ask its AI agent to videotape before the professional can say, “Stop.”

I think we as people with brain injury ought to respect the privacy of healthcare professionals and caregivers and learn to ask for permission each and every time we use AI. This respect also extends to asking permission to name them on social media, upload selfies with them, and so on.

Emancipation

For the tech-minded, open-source AI like Attie provides the possibility of creating apps for our unique needs while understanding what data is collected, how it’s used, and with whom it’s shared. Yet how can the average person keep their medical information private even when their data is not under the control of a closed ecosystem?

We can’t without awareness.

Neither can our service providers.

But organizations such as EFF (Electronic Frontier Foundation) in the US and OpenMedia in Canada can teach us how and keep us current. Brain injury organizations could use EFF’s Surveillance Self-Defense modules to guide their clients on current safe and private use of online activities and agentic AI. OpenMedia calls for plain language to foster understanding.

Healthcare providers need to create and share with clients AI policies within the broader framework of data retention and privacy practices for both staff and clients, but not in a top-down “thou shalt obey or else” tone; no adult appreciates being told off before they’ve done anything wrong or being treated like a child. Instead, approach this issue collaboratively. Collaboration breeds respect and an atmosphere of we’ll mine the best parts of AI together while protecting each other’s privacy.

One final note: People with brain injury do not require AI to write or create art. No matter how long it takes or how arduous the process, we can create beautiful works ourselves.

Copyright ©2026 Shireen Anne Jeejeebhoy

References