Adobe Illustrator CVE-2026-48275

Exploitation requires the victim to open an attacker-supplied malicious file in a vulnerable Adobe Illustrator Desktop build (2026 ≤30.5 or 2025 ≤29.8.7), and depends on the untrusted-search-path condition where Illustrator loads a component from an attacker-influenceable location (e.g., a rogue library planted in the file's directory or an earlier entry in the load search order). Limiting factors: the vector is local (AV:L), mandatory user interaction is required (UI:R) - code does not run until the victim opens the crafted file - and the attacker needs a delivery path to place both the lure file and, for search-path abuse, the malicious component where Illustrator will resolve it. There is no network-remote or unauthenticated auto-exploitation path; PR:N means the attacker needs no prior privileges on the target, but the victim's own privileges bound the resulting code execution (context of the current user).

The signals are mixed and point to a moderate, not urgent, real-world risk. The CVSS base score is high (8.6) driven largely by S:C (scope change) and total C/I/A impact, but the attack vector is local (AV:L) and requires user interaction (UI:R) to open a malicious file - so this is not remotely triggerable and not automatable. This is corroborated by CISA SSVC (Exploitation: none, Automatable: no, Technical Impact: total) and by a low EPSS of 0.16% (5th percentile). It is not listed in CISA KEV and no public exploit code has been identified. The combination of high technical impact but low exploitation likelihood and mandatory user interaction places this as a patch-in-normal-cycle item rather than an emergency, most relevant to environments where users routinely open untrusted Illustrator/design files. Note the tension worth flagging: the vector is AV:L (user must open a local file) even though the public tags label this 'RCE' - this is code execution via file-opening, not network-facing remote exploitation.

An attacker crafts a malicious Illustrator project or accompanying payload and delivers it via email, a shared drive, or a download, relying on the victim to open it in a vulnerable Illustrator build. When opened, Illustrator loads an attacker-controlled component from an untrusted search path, executing arbitrary code with the victim's privileges. No public proof-of-concept has been identified, so this scenario reflects the described mechanism rather than an observed exploit.