Disable generative AI beta in Acrobat

[howanitz]

We do not consent to giving access to our pdf contents to the generative AI beta. How do we disable this feature as an organization?

How did anyone think this was a good idea, especially considering pdf files commonly contain tax, banking, and medical information? 

I am interested in joining any upcoming privacy lawsuits.

[JR Boulay]

1.

[Lois35141403yd6c]

This answer does not stop the AI from continually prompting me to let it help. If I wanted to use an AI, I'd install one. You didn't ask; that's a problem. I can't shut it off; that's a bigger one. Please fix this.

[Caitlyn33061472mq8z]

Mine doesn't have a setting for Generative AI. 

[andreas_3904]

This setting is not available on Windows 10. This AI stuff is annoyingly slowing down text selections.

[JR Boulay]

2.

Don't be afraid.

Quote:

"We do not look at your document, prompt(s), or generated responses except in the instances described below.

Reported content, bugs, or vulnerabilities. When you report content (e.g., for being harmful, illegal, offensive, etc.), we investigate it by manually reviewing the document, prompt(s), and generated responses to make adjustments to the service to address the issue.
User-Provided Feedback. For Acrobat Individual users* that provide feedback, you have the option to share with us your document, prompt(s), and generated responses during a document session for product improvement purposes that do not include training a Large Language Model (“LLM”). Examples of product improvement include improving the operability of generative AI in Acrobat, as well as reducing hallucination, bias, and toxicity. If you do not wish to share your content, please uncheck the product improvement checkbox when you first provide feedback on a document.

If manual review of your content takes place, a limited group of trained Adobe personnel examine the content within an encrypted repository with access controls."

See:

https://helpx.adobe.com/acrobat/using/data-usage-and-handling.html

&

https://blog.adobe.com/en/publish/2024/02/20/adobes-next-generative-ai-frontier-digital-documents

We do not look at your document, prompt(s), or generated responses except in the instances described below.

• Reported content, bugs, or vulnerabilities. When you report content (e.g., for being harmful, illegal, offensive, etc.), we investigate it by manually reviewing the document, prompt(s), and generated responses to make adjustments to the service to address the issue.
• User-Provided Feedback. For Acrobat Individual users* that provide feedback, you have the option to share with us your document, prompt(s), and generated responses during a document session for product improvement purposes that do not include training a Large Language Model (“LLM”). Examples of product improvement include improving the operability of generative AI in Acrobat, as well as reducing hallucination, bias, and toxicity. If you do not wish to share your content, please uncheck the product improvement checkbox when you first provide feedback on a document.

If manual review of your content takes place, a limited group of trained Adobe personnel examine the content within an encrypted repository with access controls.

We do not look at your document, prompt(s), or generated responses except in the instances described below.

• Reported content, bugs, or vulnerabilities. When you report content (e.g., for being harmful, illegal, offensive, etc.), we investigate it by manually reviewing the document, prompt(s), and generated responses to make adjustments to the service to address the issue.
• User-Provided Feedback. For Acrobat Individual users* that provide feedback, you have the option to share with us your document, prompt(s), and generated responses during a document session for product improvement purposes that do not include training a Large Language Model (“LLM”). Examples of product improvement include improving the operability of generative AI in Acrobat, as well as reducing hallucination, bias, and toxicity. If you do not wish to share your content, please uncheck the product improvement checkbox when you first provide feedback on a document.

If manual review of your content takes place, a limited group of trained Adobe personnel examine the content within an encrypted repository with access controls.

We do not look at your document, prompt(s), or generated responses except in the instances described below.

• Reported content, bugs, or vulnerabilities. When you report content (e.g., for being harmful, illegal, offensive, etc.), we investigate it by manually reviewing the document, prompt(s), and generated responses to make adjustments to the service to address the issue.
• User-Provided Feedback. For Acrobat Individual users* that provide feedback, you have the option to share with us your document, prompt(s), and generated responses during a document session for product improvement purposes that do not include training a Large Language Model (“LLM”). Examples of product improvement include improving the operability of generative AI in Acrobat, as well as reducing hallucination, bias, and toxicity. If you do not wish to share your content, please uncheck the product improvement checkbox when you first provide feedback on a document.

If manual review of your content takes place, a limited group of trained Adobe personnel examine the content within an encrypted repository with access controls.

We do not look at your document, prompt(s), or generated responses except in the instances described below.

• Reported content, bugs, or vulnerabilities. When you report content (e.g., for being harmful, illegal, offensive, etc.), we investigate it by manually reviewing the document, prompt(s), and generated responses to make adjustments to the service to address the issue.
• User-Provided Feedback. For Acrobat Individual users* that provide feedback, you have the option to share with us your document, prompt(s), and generated responses during a document session for product improvement purposes that do not include training a Large Language Model (“LLM”). Examples of product improvement include improving the operability of generative AI in Acrobat, as well as reducing hallucination, bias, and toxicity. If you do not wish to share your content, please uncheck the product improvement checkbox when you first provide feedback on a document.

If manual review of your content takes place, a limited group of trained Adobe personnel examine the content within an encrypted repository with access controls.

We do not look at your document, prompt(s), or generated responses except in the instances described below.

• Reported content, bugs, or vulnerabilities. When you report content (e.g., for being harmful, illegal, offensive, etc.), we investigate it by manually reviewing the document, prompt(s), and generated responses to make adjustments to the service to address the issue.
• User-Provided Feedback. For Acrobat Individual users* that provide feedback, you have the option to share with us your document, prompt(s), and generated responses during a document session for product improvement purposes that do not include training a Large Language Model (“LLM”). Examples of product improvement include improving the operability of generative AI in Acrobat, as well as reducing hallucination, bias, and toxicity. If you do not wish to share your content, please uncheck the product improvement checkbox when you first provide feedback on a document.

If manual review of your content takes place, a limited group of trained Adobe personnel examine the content within an encrypted repository with access controls.

[Cyberscribe Ireland]

Wait, @JR Boulay - you start your response with "Don't be afraid" ??  For real?  This is the scariest thing that humans have invented yet to destroy themselves.  And that is the scariest answer you could give. 

[Caitlyn33061472mq8z]

AI is still a sh--tty tool and it's still Plagiarism software and it's NOT ALLOWED at my company.,

[JR Boulay]

1.

Another way is to revert to the Classic UI: https://community.adobe.com/t5/acrobat-discussions/acrobat-2023-how-to-revert-to-classic-gui-user-in...

[howanitz]

I appreciate you sharing documentation on making individual system changes. However, I still have concerns.

Adobe's documentation for turning off this feature is incorrect, quoting from the following page:

https://helpx.adobe.com/acrobat/using/generative-ai.html

=============
The generative AI features in Acrobat are turned on by default. However, you can choose to turn them off, if necessary. If you're an admin, you can revoke access to generative AI features for your team or org by contacting Adobe Customer Care. For more information, see Turn off the generative AI features.
=============

Following the link for Adobe Customer Care put in chat support, after an hour working with that system, I received the following response:

"I understand you want to disable all generative AI features in Adobe Acrobat. As checked with my resources to disable the generative AI for entire organization you'll need to contact your account representative as only they have the access to make such changes"

I then contacted CDW with this documentation and response, and they contacted Adobe as our account representative to get this change turned off. They were told that they could not turn if off for our organization either.

That leaves me doubful to the efficacy of going to each system and making Registry and PLIST changes, then reverting each system to Classic UI, since other documentation has proven to be incorrect, and there has been given no method to test. Additionally, what is to stop Adobe from pushing UI updates in the future? Do changes actually stop the upload of data, or just stop users from accessing the results of the AI?

Why should I believe Adobe has stopped data leakage of their AI, when Microsoft & ChatGPT have proven unable to do so? I am not a luddite opposed to the use of AI in many cases, I simply do not want sensitive financial and medical data shared with any AI systems, and certainly not because an employee opened a pdf file that contained such information. Apple restricts all usage of AI by employees due to data leakage, as do other tech companies, I think it might only be prudent to consider not automatically uploading content from every pdf file opened to AI as well.

https://www.wsj.com/articles/apple-restricts-use-of-chatgpt-joining-other-companies-wary-of-leaks-d4...

[Caitlyn33061472mq8z]

AND we should not have to revoke these permissions!!! They should NEVER have been enabled in the first palce!!

[Caitlyn33061472mq8z]

That does work because Adobe removed teh MENU. How convenient... Adobe used to be a great compnay that had great tool. WTF happened???

[Caitlyn33061472mq8z]

*DOES NOT WORK.

[Dan355893694p5b]

This is absurd. As admins we, and our users, were never given an opportunity to consent to this before processing of potentially sensitive data occurred. We don't list Adobe as a subprocessor of our customer's data, so it's now a compliance issue too. I don't understand how Adobe's lawyers signed off on this, it violates some pretty basic privacy requirements under a bunch of different regulations.

If anyone knows how to turn this off by policy on Mac, please let me know. On Windows, we'll be setting the bEnableAV2Enterprise registry key as suggested by https://helpx.adobe.com/acrobat/using/disable-generative-ai.html but this whole thing is a major issue. I'd also be interested if anyone has firewall rules to block this data exfiltration.

It's just absolutely mind blowing that they turned this on without any consent.

[Cyberscribe Ireland]

Howanitz - fully agree with you here, it's disgraceful and somebody should take a case.  I know this may sound ridiculous, but apart from the gross security and privacy with my clients' data, I personally feel insulted and genuinely angry every time I open my app and get prompt after prompt of "AI Assistant" offers.  I am a human being, I am educated and I am very grateful to be in a position to be able to use these amazing apps - but I do not want to hand over my precious human brain and my documents to an AI bot.  What are they thinking?

Thank you JR Boulay for providing a solution - I hope this removes these very annoying prompts as well.  HM

[Terence3575140236vn]

Adobe is unwilling to allow me, the licensed user and admin to turn off ALL ACCESS to AI.

So I think my opnly option is tocancel Adobe, and simply use MS print to PDF. A shame, because I won't beable to edit documents.

Liker, Googel, MS, and othjers, Adobe thinks they shoyuld be in control of my computer.

NO!!!!!

Time for Linux. Or better yet, no more computer

[Terence3575140236vn]

I already took Adobe off my personal machine becasue ofmAI. And Adobe has the nerve to continue to charge me for a feature I said I do not trust and need removed. If I bought something somewhere else and didn't like it I'd get my money back. What is the problem that Adobe can't just turm it off. My IT people at work don't like it, and I have already had a problem with a document because of it.

Big money, big company, small brains. Smaller ears