VOLTA virus uses AI to score victims by profit, Talos reports

The VOLTA report matters less as a story about one piece of malware than as a sign of how the economics of stolen data are changing. Traditionally an infostealer's haul is a shapeless pile that a buyer has to sort through. Talos describes VOLTA adding a scoring layer on top: addresses, employer names and receipt histories are pulled together and passed to an AI model, which sorts the victim into low, medium or high value. The detail that it falls back to another model when one refuses, up to three in all, suggests the operators treat AI refusal as a routing problem rather than a blocker.

The remaining items in the roundup point in a similar direction, toward AI becoming an everyday utility with rules attached. Anthropic renamed Claude's $200 plan from Max 20x to Ultra and removed the five-hour usage cap for verified accounts, while keeping the weekly limit in place; the $100 Max 5x becomes plain Max, with prices unchanged. OpenAI, meanwhile, is consolidating ChatGPT and Codex usage into a single allowance from December for Plus, Pro and Business plans, measured in tokens, so light chats and heavy coding compete for the same budget.

Mercari's chief AI officer described reaching 100% AI tool adoption only after the company stopped work for a day, with staff split over whether to build AI agents or hold them back. Amazon, for its part, opened early access to the generative-AI version of Alexa+ in Japan on October 29, priced at 3,000 yen a month after the early-access period, with Prime members said to pay nothing extra.