What is a zero-day exploit?
A zero-day exploit attacks a software bug before any patch exists. The vendor had zero days of warning, so there's no fix to install yet.

Zero-day exploit
A zero-day exploit is an attack on a software bug that has no fix yet. Google Threat Intelligence Group (GTIG) describes a zero-day as a flaw maliciously exploited in the wild before a patch was made publicly available. The name is about timing. The vendor had zero days of warning before attackers started using the bug.
Once the vendor ships a fix, the same bug becomes an n-day. The fix exists, but it only protects a computer where it’s actually running. A real case shows both sides. In March 2025, in a campaign Kaspersky named Operation ForumTroll, staff at media outlets, educational institutions and government organizations in Russia opened a fake invitation’s link in Chrome. That alone infected them. One of the bugs used, CVE-2025-2783 (Common Vulnerabilities and Exposures), was patched on March 25, 2025.
What a zero-day exploit actually does
Breaking into Chrome took two steps. First, code has to run in the renderer, the part of Chrome that draws pages. Kaspersky couldn’t recover the exploit used for this step. A common bug class for it is use-after-free: the browser frees a piece of memory but keeps a pointer to it. The page refills that memory with a fake object, so the stale pointer feeds the browser attacker-chosen values that steer what runs next.
The renderer is sandboxed, with almost no rights, so the attackers needed a second zero-day to break out. On Windows, -1 and -2 are pseudo-handles, shortcuts meaning the current process and the current thread. Chrome’s IPC (Inter-Process Communication) layer, Mojo, rejected -1 but not -2. The browser process passed -2 to DuplicateHandle and handed the renderer a real handle to one of its own threads. The trapped code suspended that thread, changed its registers, resumed it and ran shellcode inside the browser process.
How defenders work against zero-days
- There’s no signature yet. A brand-new exploit has no known pattern for tools to match. Defenders can’t wait for one to appear.
- Hunt for odd behavior. A browser launching
PowerShellis a strong warning sign, whatever bug caused it. - Watch for persistence keys. A new COM (Component Object Model) registry key, such as an override under
HKCU\Software\Classes\CLSID, can make Windows reload malware. - Patch and relaunch. Once the fix ships, the zero-day becomes an n-day. Chrome’s update only takes effect after you close and reopen the browser.
The result: behavior hunting can catch the attack while no fix exists, and a running patch closes that hole.
A zero-day is not a missed update
Many people think a fully updated computer is safe from a single click. A zero-day breaks that idea. It’s used before any patch exists, so there’s nothing to install yet. The ForumTroll victims needed no further action after opening the link. The opposite trap comes later. When a patch ships, it only stops the attack once it’s actually running, and Chrome’s fix only runs after a relaunch.
What to do
- Open Chrome’s menu, then Help, then About Google Chrome, and click Relaunch if it’s there.
- Close and reopen your browser whenever an update is waiting, because a downloaded fix isn’t a running fix.
- Treat unexpected invitations with care, since opening one link was enough in this attack.
- If you defend a network, alert on a browser launching
PowerShelland on new COM registry keys that load code.
Watch it in 60 seconds
