Random false positive gemini filters

Google Antigravity
,
1

I am analysing a binary and trying to see what it does but gemini constently rejecting my every promt even randoms

image
image884×270 5.31 KB

if the case of reverse enginnering is violated please tell em and I will just end my subs. But I didnt see anything about reverse engineering violancy on your policies?

2

never got this problem before btw, I am getting those filter blocks liek 1-1.5 days or sth

3

Hi @swansizz,

Welcome to the Fourm, Thanks for reaching out.

To give you some context on what’s happening: Gemini’s safety filters evaluate the entire conversation history sent with each turn. When conducting binary analysis or reverse engineering, decompiled code or raw assembly snippets in earlier turns can sometimes trigger an automated safety filter false positive. Because that flagged context remains in the active chat thread, every subsequent prompt (even simple inputs like ‘ee’) ends up getting blocked too.

To resolve this issue and get unblocked, you can try these steps:

  1. Start a New Chat Session: Click + New Chat'(or press Ctrl+N / Cmd+N) to clear the accumulated conversation history that contains the flagged binary context.
  2. Frame Binary Context: When sharing assembly or decompiled code, try including explicit context in your prompt indicating that you are performing defensive security research, debugging or educational software analysis.
  3. Report False Positives: Click the inline send feedback link directly on the filter block message. This submits the prompt turn to our safety engineering team so we can refine our classifier boundaries for security research and reverse engineering workflows.
4

Thank you after new chat even 1 or 2 message getting flagged. I dont know I am just continuing my old project and never got this error before its weird.

5

nope still same, btw where can I report the report section redirecting me here