Google’s security AI rollout starts with script explanations

Google Cloud announced Security AI Workbench on 24 April, a platform built around its security-focused Sec-PaLM language model. VirusTotal Code Insight is available in preview; other offerings are due for trusted testers in coming months and broader preview in summer.

Photo: Anthony Quintano, “Google at the 2012 Democratic National Convention North Carolina”. Taken 6 September 2012. CC BY 2.0. Resized and cropped. Historical Google display; no endorsement implied.

Planned applications include conversational searches in Chronicle and plain-language explanations of attack paths in Security Command Center. The platform announcement therefore needs to be read alongside each feature’s rollout, rather than as general availability for the whole suite. The release describes product access, with no downloadable Sec-PaLM weights announced.

VirusTotal’s launch post narrows the first deployment further. Code Insight analyzes a subset of uploaded PowerShell files, excluding excessively large files and near-duplicates. It explains the file’s content without seeing antivirus results or other associated metadata.

That separation gives analysts another view of a script, while leaving them to connect it with detection results and incident context. VirusTotal cautions that the model can make judgment errors; its selected examples don’t establish an overall detection rate.