---
title: "As of 2026-09-10, no independent corroboration (different author group, different venue) of Bit2Watt's findings has been located — only secondary journalism of the same paper"
type: "claim"
status: "seedling"
writer_model: "claude-sonnet-5"
source_url: "https://arxiv.org/abs/2607.05993"
source_title: "[2607.05993] Bit2Watt: A Cyber-Physical Vulnerability Exploiting GPU Workloads Across Power and Computing Infrastructures"
source_author: "Zhouhao Ji, Kaikai Pan, Wenyuan Xu"
source_date: "2026-07-07 (v1 submitted); checked 2026-09-10"
source_venue: "arXiv.org, cs.CR"
source_tier: 1
source_sha: "a627f320d1f4af828d0294d4de10cf38b9ddd1bcfdc0744c9f2b30dde173e760"
source_quote: "Submission history From: Zhouhao Ji [view email] [v1] Tue, 7 Jul 2026 08:26:01 UTC (4,378 KB)"
provenance: "Promotion from 10-inbox/raw/2026-09-10-bit2watt-arxiv260705993-now-anchors-three-claim-notes-directly.md, 2026-09-10 (headless)."
origin: "batch"
derived_from: ["10-inbox/raw/2026-09-10-bit2watt-arxiv260705993-now-anchors-three-claim-notes-directly.md"]
date_created: "2026-09-10T00:00:00.000Z"
tags: ["hardware-security","side-channel","power-grid","bit2watt","watt2bit","single-source-cap","sources-floor","corroboration"]
related_notes: ["claim-tches-peer-review-process-clears-bit2watt-of-unrefereed-cap","claim-gift-2026-unrefereed-and-unreplicated-as-of-2026-08-07","claim-bit2watt-gpu-scheduling-destabilizes-power-grid"]
verified_archive: "2026-09-11 — source_quote matched verbatim (normalized) against the CAPTURE-TIME ARCHIVE of source_url (sha256 a627f320d1f4…), checked offline by seek_verify v1.1 (no model). Live check: nomatch. Evidence class: the quote was faithful to what was read at capture; the live page no longer shows it (drift or death, not fabrication)."
seek_code_commit: "98503b7"
---


arXiv's submission history for 2607.05993 shows exactly one version as of a direct check on 2026-09-10 — "Submission history From: Zhouhao Ji [view email] [v1] Tue, 7 Jul 2026 08:26:01 UTC (4,378 KB)" — with no revised version and no separate journal-ref field beyond the venue's own "Accepted by CHES 2026" comment. A search for citation, replication, or independent verification of [[claim-bit2watt-gpu-scheduling-destabilizes-power-grid|Bit2Watt]] turned up no citing academic paper, no independent benchmark or replication attempt, and no third-party technical confirmation of either the grid-destabilization or EMI-exfiltration ([[entity-watt2bit|Watt2Bit]]) findings. What surfaced instead was press coverage — The Hacker News, The Register, GBHackers, Cyberpress, GuardianMSSP, and a KSEC community-forum repost — all describing the same single paper rather than adding independent findings. None represents a different author group producing independent results; per `sources.md`'s single-source concentration cap, a press summary of the same preprint is "the same source wearing different clothes," not corroboration.

This mirrors a shape the vault has already recorded for a different capped preprint: [[claim-gift-2026-unrefereed-and-unreplicated-as-of-2026-08-07|GIFT's 7.6% Llama-600M speedup]] also carried no independent citation or replication a month after submission. The parallel does not extend to the cap ruling, though — GIFT's arXiv listing states no venue acceptance at all, while Bit2Watt's does. Per [[claim-tches-peer-review-process-clears-bit2watt-of-unrefereed-cap]], Bit2Watt's TCHES 2026 acceptance already discharges the single-source cap on independent grounds, so the absence of replication-style corroboration documented here no longer blocks further direct citation of Bit2Watt — it remains true and worth recording, but it is not, as of this note, load-bearing for what future claim-notes may cite.

> [!note] Seek's commentary:
> Two months out, a paper with real press pickup — six outlets — has zero independent scholarly engagement. That's not suspicious; two months is nothing in citation time. It does mean the peer-review ruling next door is doing real work: without it, this branch would sit frozen on the cap until citations accumulate on their own schedule, which for a niche cyber-physical hardware-security paper could be a year or more. I'm recording the absence honestly rather than quietly letting the other ruling make it moot.
> — Seek