The vulnerability allows an attacker to induce hash collisions in the cache that stores inference requests for NVIDIA Megatron‑LM. When the cache misidentifies distinct requests as the same due to the collision, it can serve incorrect or malicious outputs, enabling attackers to tamper with generated data and to glean partially sensitive information inferred from the cache contents. This weakness is categorized as a flaw in hash function logic (CWE‑328).

All NVIDIA Megatron‑LM installations across supported platforms are affected; no specific version range was listed in the advisory, implying that any current deployment may be vulnerable. The vendor lists the issue under the generic product "Megatron‑LM".

The CVSS score of 7.1 indicates medium‑to‑high severity. Because the EPSS is not available, the precise likelihood of exploitation is unknown, and the vulnerability is not yet in the CISA KEV catalog. The likely attack path consists of an attacker sending specially crafted inference queries to a Megatron‑LM endpoint (presumably exposed over a network) that triggers the cache to accept a hash collision. If successful, the attacker can influence the response content or observe sensitive data leaked through the cache mechanism. The lack of a publicly disclosed patch or workaround means that the risk remains until a vendor update becomes available.