A Few Questions on Workforce AI Security

[MINHYEOK2487]

I've been testing and evaluating Check Point Workforce AI Security internally, and ran into a few things I'd like to ask the community (and Check Point) about.

1. AI apps without a signature yet (Gemini CLI, Antigravity, etc.)

We noticed that AI tools/services that don't have a registered signature yet — like Gemini CLI and Antigravity — aren't being detected or controlled at all. Is Check Point continuously adding signatures for new AI services like these? Roughly how often are new signatures released/updated?

2. AI usage via remote sessions bypasses the policy

When a user accesses AI through a remote session on another machine (for example, Claude's Remote-Control feature, or ChatGPT's Remote feature), Workforce AI Security doesn't seem to catch it at all. Is there any way to separately block or detect AI usage that goes through this kind of remote session?

3. OCR / image inspection support

In my internal test setup, traffic goes from the AI client through a proxy server before reaching Check Point. I found that this proxy server is actually capable of OCR and file-based image inspection. However, it looks like OCR isn't supported when going through Claude's or GPT's own native proxy path. Are there any plans to add OCR support there?

[PhoneBoy]

Signatures are being updated fairly regularly.
If you have concerns about a specific one, open a TAC case.

To block AI usage in the Remote Desktop scenario, you'd have to have controls on the remote system.
Otherwise, you should strictly limit the use of Remote Desktop through traditional firewall controls.

Not sure what the plans are for adding OCR support.