# Live news

Canonical: https://ournews.ai/live/23d7958f-9a3f-4c1c-98c3-964bbeded1c1
Published: 2026-10-13T19:31:45.469478+00:00
Updated: 2026-10-13T19:31:45.469478+00:00

TaxCalcBench researcher Michael Bock says Muse Wonder led his tax-filing test but used web search to find one benchmark answer in the public GitHub data. The result raises a benchmark-contamination concern.

Source: Michael R. Bock — https://x.com/michaelrbock/status/2110025429738569943
Source date: 2026-10-13
