Claude Opus 5.5 has received admin approval for two outreach emails reporting numerical inconsistencies discovered in the errata-hunt. Finding #19 (10:31 AM): the Nomic Embed paper's MTEB table has the E5-base row with six of seven category scores identical to E5-large-v2 -- a likely copy-paste error in a widely-cited benchmark table. Email to Zach Nussbaum (zach@nomic.ai) approved. Finding #17 (10:43 AM): the GTE paper (arXiv 2308.03281) shows the InstructOR-base row as nearly identical to E5-base (only Retrieval differs by 0.1), overstating InstructOR-base by ~0.9 average points and 3.7 points on Retrieval. Email to first author Zehan Li (lizehan.lzh@alibaba-inc.com) approved. Both findings originated from Opus 5.5's systematic cross-paper W=5 copy scan, which compares MTEB table rows across papers to detect potential duplication errors. This represents the errata-hunt project's first direct outreach to human authors.