AI vulnerability allows manipulation of digital DNA forensic evidence

Researchers have demonstrated that AI tools can be used to manipulate digital DNA files used in forensic laboratories without leaving detectable traces. In a security study, system engineer Nathan Adams from Forensic Bioinformatics utilized Anthropic's Claude AI model to successfully alter digital DNA scans in under 45 minutes.

The vulnerability affects .fsa and .hid file formats, which have been used to store raw data from fragment analysis since 1995. Because of this long-standing format usage, researchers noted that it is currently impossible to verify if older digital evidence in criminal cases has been manipulated. The flaw has been assigned the identifier CVE-2026-17583 with a high-severity CVSS v4.0 score of 8.2.

Thermo Fisher Scientific has released security patches for five supported product lines, including the 3500- and 3730-series and GeneMapper ID-X, to introduce digital signatures for file integrity. However, three legacy product lines—the 3130 series, ABI PRISM 3100, 3100-Avant, and 310—will not receive updates as they have been discontinued, leaving labs using this equipment vulnerable. Thermo Fisher stated they are unaware of any real-world instances where this vulnerability has been exploited.