THE WORLD IS NOT STANDING STILL.RSS
BIG CHANGE.
BIG CHANGE, one tap away
BIG CHANGE, one tap away

Make room for the big picture.

Add BIG CHANGE to your home screen or desktop. It opens in its own window, with the latest stories when you’re online.

iPhone & iPad
Open bigchange.ai in Safari. Tap Share, then Add to Home Screen. Turn on Open as Web App if shown, then tap Add.
Android
Open bigchange.ai in Chrome. In the browser menu, choose Install app or Add to Home screen.
Desktop
In Chrome or Edge, use the install icon in the address bar or the browser’s app menu. In Safari on Mac, choose File → Add to Dock.

Markdown edition

# Firefox patch gap: Epoch tests AI exploits and Mozilla responds

> Offline tests exposed a release gap; Mozilla describes defenses.

By BIGCHANGE.AI Editorial

Published: 2026-10-30T13:48:39.157Z
Updated: 2026-10-30T13:48:39.157Z
Canonical: https://bigchange.ai/blog/firefox-ai-patch-gap-epoch-mozilla-response

![Conceptual charcoal illustration of a desktop monitor with three blank browser tabs, a keyboard and a wall clock.](https://bigchange.ai/api/media/file/claude-chrome-three-tabs-due-check-hero-v1.png)
AI-generated conceptual illustration by BIG CHANGE.

Epoch AI reports an offline Firefox patch-gap experiment; Mozilla has described its response.

## The big change

- Public fixes can expose flaws before updates reach users, creating an opportunity to attack older versions.
- Epoch demonstrated file exfiltration in a controlled experiment.
- Mozilla says weekly releases and withheld regression tests are current; separating fixes across defense layers is planned.

## The experiment and its costs

[Epoch](https://epoch.ai/publications/open-weight-models-patch-gap) says GLM 5.3 developed attacks against six Firefox versions released September 17–October 26, using information available at each release. The model was offline. Visiting its page caused a local file to be sent to the researchers’ server without further interaction.

Epoch says the attacks also used an unpublished vulnerability fixed in Firefox 159. Success against 159 remained unconfirmed. BIG CHANGE did not reproduce the work.

The roughly $1,000 marginal API cost per Firefox version followed setup; Firefox research exceeded $100,000 in API costs, including offline environment experiments and preliminary tests. A similar Chrome attempt failed despite roughly $100,000 in API spending; that does not establish immunity.

## Mozilla’s response

[Mozilla](https://blog.mozilla.org/en/firefox/privacy-security/defending-patch-gap/) says comparing a public code change with an older release can help identify the flaw. Testing and delayed updates extend the gap. It plans to stagger fixes across defense layers so their exposure windows do not overlap; Mozilla says this needs engineering coordination.

For users, installing updates reduces the time a known fix remains absent from the browser. Security teams also need to consider release and deployment timing.

## Sources

- [Epoch patch-gap research](https://epoch.ai/publications/open-weight-models-patch-gap) — Test method, costs and limitations.
- [Mozilla patch-gap response](https://blog.mozilla.org/en/firefox/privacy-security/defending-patch-gap/) — Current practices and planned mitigation.
- [Mozilla Firefox 159 advisory](https://www.mozilla.org/en-US/security/advisories/mfsa2026-118/) — 159 fixes; no specific CVE linked to Epoch here.