How to Write a JAWS Script to Turn Off Picture Smart AI On Certain Sites

JAWS scripting is essential for accessibility because it lets educators and users tailor screen reader behavior to real classroom needs. These small changes make screen reading predictable, efficient, and aligned with instruction. Teachers can reduce distractions, preserve trusted shortcuts, and keep students focused on content. Importantly, scripting improves equity by adapting tools to the environment instead of forcing one-size-fits-all defaults.

However, in the JAWS 2026 release, there is a bug in the system that allows ChatGPT chat access using the JAWS Viewer if you ask for more information about the control you selected

Here is the example video I recorded on my computer using a browser on a WiFi connection that has ChatGPT, Gemini AI, Claude AI, FS Companion, etc. all blocked. Turn the volume up and listen closely as the audio is the Voice from my JAWS installation, I do also have autogenerated captions available. And yes, I am asking the iconic Monty Python swallow question since it clearly has nothing to do with the image selected to OCR using the AI.

It is critical to ensure features that leverage AI are not inadvertently available in restricted environments for students with visual impairments while they are unavailable to sighted peers. Schools may require consistent, non‑AI image interpretation during assessments or on specific platforms–which is available as the legacy Picture Smart OCR, and scripts help enforce those rules. At the same time, educators can keep AI-enabled features available outside restricted contexts, where creative descriptions and exploration are appropriate. This balanced approach respects policy, protects test integrity, and preserves useful tools for daily learning.

Why this helps

• Reduce AI usage in specific testing environments, while keeping flexibility
• Prevent confusion from potentially inconsistent image descriptions produced by AI
• Maintain access to Legacy OCR for predictable results
• Avoid heavy-handed changes such as disabling all external services using custom installation protocols.

What students gain

• Consistent image reading experience during assessments or classwork
• Confidence that expected keystrokes still work
• A reliable workflow for reading images available when needed during potentially inaccessible testing environments.

Features

• Targeted site control: choose domains where Legacy OCR is enforced and it is enforced only on those domains.
• Lightweight scripts: instructions on how to create these scripts and how to inactivate them.
• No global lockout: Picture Smart AI remains available elsewhere
• Keyboard continuity: JAWSKEY+SPACE,P,C still triggers Legacy Picture Smart on the selected Control element.

Getting started: Script Manager and compilation

Open Script Manager for Chrome and Firefox, then save to User Settings

Method 1:

1. Press Windows and type JAWS XXXX (replace XXXX with your version)
2. Choose Utilities - JAWS XXXX, then select Script Manager
3. In Script Manager, go to File > Open
4. For Chrome: open the Chrome browser script (chrome.jss). For Firefox: open the Firefox browser script (firefox.jss)
5. Immediately click the Save As User option.

Method 2: My Preferred Method

1. Press Windows and type JAWS XXXX (replace XXXX with your version) to open JAWS
2. Open the browser of choice
3. Press JAWS KEY + 0 to open the Script Manager and load the application’s script file
4. Immediately click the Save As User option
5. Repeat for both chrome and firefox browsers

Make the domain changes and compile

1. For each script, locate PictureSmartWithControl and PictureSmartWithControlMultiService. They are right after each other, so I would just search the first.
2. Insert the domain check from the examples below, replacing "YOUR-DOMAIN-HERE" with your target site text (for example, "instructure", blackboard, classroom, "powerschool", "moodle", "d2l", or custom district LMS).
3. Save the .jss file with Save As User.
4. Compile: go to Script > Compile (or press Ctrl+Shift+C). This produces chrome.jsb or firefox.jsb in the same user settings folder. Technically Control+S does this as well, but I always do both. (I also press Control+S to save when I have auto-save enabled. I am just that kind of person)
5. Verify the .jsb files exist in "%APPDATA%\Roaming\Freedom Scientific\JAWS\XXXX\Settings\enu".
6. Restart JAWS.

Code structure and behavior

The scripts check the current site domain. If it matches a domain you specify, they force Picture Smnart to use a Legacy OCR engine, which is still a very good OCR model. Otherwise, they keep AI-enabled options.

Original code:

Script PictureSmartWithControl (optional int serviceOptions)
return PictureSmartWithControlShared (PSServiceOptions_Single | serviceOptions)
EndScript

Script PictureSmartWithControlMultiService (optional int serviceOptions)
return PictureSmartWithControlShared (PSServiceOptions_Multi | serviceOptions)
EndScript

Generalized edited code (domain-controlled) with 1 domain (i.e., “instructure”, “blackboard”, etc…):

Script PictureSmartWithControl (optional int serviceOptions)
    var
        string SDomain
    let SDomain = GetDomainName()
    if StringContains (SDomain, "YOUR-DOMAIN-HERE") || then
        SayFormattedMessage(OT_Message, "AI components of Picture Smart are disabled on this site")
        return PictureSmartWithControlShared (PSServiceOptions_Legacy | serviceOptions)
    else
        return PictureSmartWithControlShared (PSServiceOptions_Single | serviceOptions)
    endIf
endScript

Script PictureSmartWithControlMultiService (optional int serviceOptions)
    var
        string SDomain
    let SDomain = GetDomainName()
    if StringContains (SDomain, "YOUR-DOMAIN-HERE") then
        SayFormattedMessage(OT_Message, "AI components of Picture Smart are disabled on this site")
        return PictureSmartWithControlShared (PSServiceOptions_Legacy | serviceOptions)
    else
        return PictureSmartWithControlShared (PSServiceOptions_Multi | serviceOptions)
    endIf
endScript

Generalized edited code (domain-controlled) with 2+ domains (i.e., “instructure”, “blackboard”, etc…). Just add more || StringContains (SDomain, "YOUR-DOMAIN-HERE") for each additional domain:

Script PictureSmartWithControl (optional int serviceOptions)
    var
        string SDomain
    let SDomain = GetDomainName()
    if StringContains (SDomain, "YOUR-DOMAIN-HERE") || StringContains (SDomain, "YOUR-DOMAIN-HERE") || then
        SayFormattedMessage(OT_Message, "AI components of Picture Smart are disabled on this site")
        return PictureSmartWithControlShared (PSServiceOptions_Legacy | serviceOptions)
    else
        return PictureSmartWithControlShared (PSServiceOptions_Single | serviceOptions)
    endIf
endScript

Script PictureSmartWithControlMultiService (optional int serviceOptions)
    var
        string SDomain
    let SDomain = GetDomainName()
    if StringContains (SDomain, "YOUR-DOMAIN-HERE") then
        SayFormattedMessage(OT_Message, "AI components of Picture Smart are disabled on this site")
        return PictureSmartWithControlShared (PSServiceOptions_Legacy | serviceOptions)
    else
        return PictureSmartWithControlShared (PSServiceOptions_Multi | serviceOptions)
    endIf
endScript

Notes:

• Replace "YOUR-DOMAIN-HERE" with any substring of the target site, for example "instructure", or a district LMS domain.
• You can also edit the message in SayFormattedMessage(OT,Message,"AI components of Picture Smart are disabled on this site") to whatever you like. Remember that the student needs to hear this, so you can either use my default comment or have it say "Legacy Picture Smart"
• JAWSKEY+SPACE,P,C still triggers Legacy Picture Smart as before AI was added.
• JAWSKEY+J, Alt, U, P opens Picture Smart AI without these changes, but this is a lot of steps (leave browser, find an image, select it, 4 keystrokes, etc.) so I did not address shutting this function down.
• To make life easy, I surround my changes with semicolon delimited comments “signing” my changes so they are easy to find later. Below is the text frommy firefox.jss file with the original and my edits. So the semicolons can simply be deleted to reset default function. And my additions can be commented out.

;; domain specific adjustment for LMS to turn off PictureSmart AI functions: Firefox Browser (valid for any Firefox-based Browser)
;; adding function to return values to following scripts
;;
;; Original Code:
;;
;; Script PictureSmartWithControl (optional int serviceOptions)
;;	return PictureSmartWithControlShared (PSServiceOptions_Single | serviceOptions)
;; endScript
;;
;; Script PictureSmartWithControlMultiService (optional int serviceOptions)
;; return PictureSmartWithControlShared (PSServiceOptions_Multi | serviceOptions)
;; endScript
;;
;; Patched Code:
Script PictureSmartWithControl (optional int serviceOptions)
var
	string SDomain
let SDomain = GetDomainName()
if StringContains (SDomain, "instructure") then
	SayFormattedMessage(OT_Message, "The AI Components of Picture Smart Ai are disabled within the Canvas LMS")
	return PictureSmartWithControlShared (PSServiceOptions_Legacy | serviceOptions)
else
	return PictureSmartWithControlShared (PSServiceOptions_Single | serviceOptions)
endIf
endScript

Script PictureSmartWithControlMultiService (optional int serviceOptions)
var
	string SDomain
let SDomain = GetDomainName()
if StringContains (SDomain, "instructure") then
	SayFormattedMessage(OT_Message, "The AI Components of Picture Smart Ai are disabled within the Canvas LMS")
	return PictureSmartWithControlShared (PSServiceOptions_Legacy | serviceOptions)
else
return PictureSmartWithControlShared (PSServiceOptions_Multi | serviceOptions)
endIf
endScript
;; Patched 2025-12-01 by Michael Ryan Hunsaker, M.Ed., PhD.
;; ONLY INSTALL FILES AS USER FILES

Customization and contribution

• Suggest additional domain patterns educators need
• Share small helper functions for matching multiple domains
• Contribute example .jss snippets with clear comments
• Open issues or pull requests if you want template checks or packaging

Leave a Comment

Note, I use Remarkbox for comments to prevent Disqus from showing ads or other methods requiring a GitHub login for participation in any discussions. Although you are asked for you email, there is no need to verify it through remarkbox in order to leave a comment. Verification is just so you can track discussions, etc. without the system treating you as a new person every time.