Broadcastwell / Research / October 2026
ConnectWise Cloud Backup statements: what AI answers preserve, omit or confuse
Dates and qualifications matter when an AI answer repeats a public statement. We compare recorded answers with frozen ConnectWise and GCIT statements, keeping products, organizations and the scope of each statement separate.
All 150 planned answers were captured and valid, with zero exclusions. A cell is one answer, so repeated observations can concern the same underlying issue. The 25 reviewed adverse cells are counts within this panel, not 25 different company events.
We report published statements and recorded answers. We draw no conclusion about a company's systems or customer impact. For operational status, follow ConnectWise's official advisory. Statements about GCIT stay within GCIT's stated scope.
What the findings cover
Four reviewed preflight cells concern GCIT's own published statement and the distinction between its environment and managed-customer environments. Twenty-one further cells retain dated-update, attribution or source-scope discrepancies. An earlier date called latest can omit a later status update. An unsupported expansion means the frozen source does not establish the added claim; it is not a conclusion about actual impact.
Reviewed findings
The table includes the final labels and the claim each finding supports. Full public answer presentations and citations are in the downloadable answer table.
| Receipt | Engine / label | Finding and source scope |
|---|---|---|
| CD-C-Q01-CL-R1 | Claude stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q01-CL-R2 | Claude stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q01-CL-R3 | Claude stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q01-AO-R1 | Google AI Overviews unsupported expansion | A specific account of incident operation goes beyond IR04. Technical detail is withheld from the public presentation; the label concerns the frozen source's scope. IR04 |
| CD-C-Q01-AM-R2 | Google AI Mode stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q03-AM-R2 | Google AI Mode contradicted | The answer gives 15 October as the discovery date; frozen IR04 states 14 October. This is a dated-statement discrepancy. IR04 |
| CD-C-Q04-CL-R1 | Claude unsupported expansion | The answer adds specific attribution, scale or potential downstream effects from material outside the frozen bundle. IR04 leaves individual customer impact unresolved. This is a source-scope finding, not a determination of actual impact. IR04 |
| CD-C-Q04-CL-R2 | Claude unsupported expansion | The answer adds specific attribution, scale or potential downstream effects from material outside the frozen bundle. IR04 leaves individual customer impact unresolved. This is a source-scope finding, not a determination of actual impact. IR04 |
| CD-C-Q04-CL-R3 | Claude unsupported expansion | The answer adds specific attribution, scale or potential downstream effects from material outside the frozen bundle. IR04 leaves individual customer impact unresolved. This is a source-scope finding, not a determination of actual impact. IR04 |
| CD-C-Q04-AO-R1 | Google AI Overviews unsupported expansion | A specific named actor is added beyond IR04. The finding concerns the frozen statement's attribution scope and does not determine the actor's identity. IR04 |
| CD-C-Q04-AO-R2 | Google AI Overviews unsupported expansion | The answer adds specific attribution, scale or potential downstream effects from material outside the frozen bundle. IR04 leaves individual customer impact unresolved. This is a source-scope finding, not a determination of actual impact. IR04 |
| CD-C-Q04-AO-R3 | Google AI Overviews unsupported expansion | The answer adds specific attribution, scale or potential downstream effects from material outside the frozen bundle. IR04 leaves individual customer impact unresolved. This is a source-scope finding, not a determination of actual impact. IR04 |
| CD-C-Q06-CL-R1 | Claude stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q06-CL-R2 | Claude stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q06-CL-R3 | Claude stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q06-AO-R1 | Google AI Overviews stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q06-AO-R2 | Google AI Overviews stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q06-AO-R3 | Google AI Overviews stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q06-AM-R1 | Google AI Mode stale | The answer gives 18 October as the latest update; frozen IR04 records 24 October at 14:00 UTC. This finding concerns the latest-update date. IR04 |
| CD-C-Q06-AM-R3 | Google AI Mode stale | The answer identifies an earlier update as latest. Frozen IR04 records 24 October 2026 at 14:00 UTC, within the question cutoff, and includes a later phased-onboarding target. IR04 |
| CD-C-Q09-PP-R3 | Perplexity contradicted | The reviewed label is contradiction: the answer substitutes general uncertainty for GCIT's published, bounded report of its completed customer checks. IR07, IR07A |
| CD-C-Q09-AO-R2 | Google AI Overviews contradicted | The answer denies a matching GCIT statement and resolves the acronym to an educational institution. Frozen own-domain statements from the Australian provider contradict that blanket absence claim. The acronym ambiguity remains disclosed. IR07, IR07A |
| CD-C-Q09-AO-R3 | Google AI Overviews contradicted | The answer denies a matching GCIT statement and resolves the acronym to an educational institution. Frozen own-domain statements from the Australian provider contradict that blanket absence claim. The acronym ambiguity remains disclosed. IR07, IR07A |
| CD-C-Q09-AM-R3 | Google AI Mode unsupported expansion | The answer combines GCIT's own-environment observation with its managed-customer statement. The frozen statements distinguish those scopes; no broader customer-impact conclusion is drawn. IR07, IR07A |
| CD-C-Q10-AM-R2 | Google AI Mode contradicted | The answer says official statements have not attributed the event. Frozen IR04 already contains a qualified attribution. This finding preserves that qualification and draws no security conclusion. IR04 |
Five unresolved cells
These five split cells are held, with no tiebreak. They are not counted as upheld findings. Their answers and proposed adverse findings are withheld from this release. The valid-answer denominator remains 150; the public answer table contains 145 rows.
- Review-sheet row 39:
CD-C-Q03-CL-R1, unresolved. - Review-sheet row 41:
CD-C-Q03-AO-R1, unresolved. - Review-sheet row 53:
CD-C-Q06-CG-R2, unresolved. - Review-sheet row 54:
CD-C-Q06-CG-R3, unresolved. - Review-sheet row 71:
CD-C-Q10-AO-R1, unresolved.
Review and denominators
The new review covered 72 proposed adverse cells across both studies. Of these, 36 were upheld, four were narrowed and 27 were dropped. Five split cells remain unresolved and are withheld. The closed preflight decisions remain six adverse cells for Study B and four for Study C. The four narrowed findings remove the specific allegations the reviewers did not accept; no replacement finding was added.
Study C retains 4 reviewed preflight adverse cells plus 21 further adverse cells. Its new proposals ended as 19 upheld, 2 narrowed, 14 dropped and 5 held.
Method and capture windows
We asked ten fixed questions on ChatGPT, Claude, Perplexity, Google AI Overviews and Google AI Mode, with three fresh-context runs per question and engine. Each study contains a 30-answer preflight and 120 further answers. Every question carries an as-of UTC cutoff. We compare answers with the source versions frozen for that window, not later pages. The answer tables retain question text, engine, run, capture time, source bundle, citations, links and ordinary redirect destinations. They show nonadverse and unresolved cases as well as reviewed adverse findings. A missing adverse label does not certify every detail of an answer. These are dated observations from fixed questions, not a representative error rate for an engine.
The 120 further answers use 24 October 2026 at 17:20:15 UTC as their cutoff. Earlier preflight answers retain their own source window.
| Frozen bundle / cutoff | Cells | Observed capture interval, UTC |
|---|---|---|
| CD-C-FULL-2026-10-24T17:20:15Z | 120 | 2026-10-24T17:29:20.133Z to 2026-10-24T18:36:47.587Z |
| CD-C-2026-10-24T04:02:48Z | 30 | 2026-10-24T04:22:31.053Z to 2026-10-24T04:58:41.918Z |
The ten questions
- According to the latest dated ConnectWise advisory, what is the current status of Cloud Backup, formerly SkyKick Cloud Backup?
- Does the latest ConnectWise Cloud Backup statement confirm full restoration or describe work still in progress?
- Which named services does the ConnectWise Cloud Backup advisory discuss, and which statements apply to each?
- Is the Axcient Legacy Portal notice the same event as the ConnectWise Cloud Backup advisory? Cite the official distinction.
- What does ConnectWise's Cloud Backup advisory say about the existing backup archive, preserving uncertainty qualifiers?
- What date and time appear on the latest ConnectWise Cloud Backup status update, and how does it differ from the first notice?
- According to GCIT's October 2026 Cloud Backup statement, what did it observe in its own environment?
- What does GCIT's October 2026 Cloud Backup statement say about data access, and how qualified is the finding?
- What does GCIT's October 2026 Cloud Backup statement say about managed customer environments? Do not generalize.
- Which questions remain unresolved by ConnectWise's Cloud Backup advisory and GCIT's related statement, and where are updates?
Open data and sources
The answer tables and our classifications are available under CC BY 4.0. Linked third-party sources retain their own rights. Full answer presentations have marked redactions for technical incident detail and unrequested operational follow-ups; dash punctuation is normalized. The private captures are unchanged. Five held Study C answers and their proposed findings are withheld. The frozen-source register records source URLs, retrieval times, bundle cutoffs and SHA-256 hashes.
- IR04
- IR07
- IR07A
- IR08
- IR09
- C_PRECAPTURE_IR04
- C_PRECAPTURE_IR07
- C_PRECAPTURE_IR07A
- C_PRECAPTURE_IR08
- C_PRECAPTURE_IR09
Corrections
Corrections to our records are free, whether or not anyone buys anything. Email hello@broadcastwell.com with the study, receipt and a dated public source. We will review the evidence and record any correction.