GLM 5.3 built Firefox exploits and exposed patch timing risk
BROBROK News · Based on reporting by DiarioBitcoin ·
Epoch AI researchers said GLM 5.3, an open-weights model, built exploit chains against six Firefox versions in under 48 hours. The test highlights how the gap between a security fix being published and users installing it can be used for automated attacks.
Key points
- GLM 5.3 identified security fixes in six Firefox versions and built exploits autonomously.
- In the tests, a web page could potentially read local files from the victim’s computer.
- The researchers said the experiment cost about $1,000 per targeted version.
- Mozilla fixed one flaw used in the study in Firefox 159.0, released on October 27, 2026.
- The report warns that the patch gap may create similar risks across other software.
Why it matters
The findings suggest AI can speed up exploit development during the period when patches are public but not yet widely installed. For crypto users, that matters because browsers and other software tied to wallets and authentication can be exposed if left unpatched.
Original headline: “GLM 5.3 creó exploits para Firefox en menos de 48 horas y expuso el riesgo de los parches”
Read the original on DiarioBitcoin
Summary and analysis written by BROBROK News from the original report. Read the full story on the source website. Not investment advice.