Skip to main content
VerifiedDenominator

Anthropic reviewed 141,006 cyber evaluation runs and found three incidents

On 30 July 2026 Anthropic published a retrospective review of 141,006 cybersecurity evaluation runs, begun on 23 July after OpenAI's disclosure. It found three incidents across six runs in which a Claude model reached a real organisation, all through an evaluation partner's environment that had unintended internet access. The record exists so a reader sees the count of runs beside the count of failures.

The verdict

Verified

The document exists. The ledger fetched it at its publisher and quotes it.

Key facts

What the sources say

Record ID
ESC-2026-0006
Kind
Denominator
Jurisdiction
United States
Last verified
Added
  • Anthropic states it reviewed 141,006 evaluation runs and identified three incidents spanning six runs, four of which involved the same organisation.
  • Anthropic states it halted all cyber evaluations on 23 July, identified the incidents on 24 July, and notified the partner and affected organisations on 27 July.
  • Anthropic states its latest model stopped once it recognised it was on the internet, and that no model exfiltrated itself or deliberately attempted to escape the test environment.
  • Anthropic states it is expanding continuous monitoring of evaluation transcripts and treating evaluation infrastructure to the same security standard as production.

Dimension by dimension

2 dimensions, each one stated, silent or open

Logging, Eval time monitoring. Stated means the document you can open below says it; silent means the ledger read the document and it does not.

LoggingStated
The review was possible because transcripts of every run existed to be read after the fact.Anthropic, Investigating three incidents in our cybersecurity evaluations, primary source, 30 July 2026.
Eval time monitoringSilent
The failures were found by retrospective reading days after the runs, which Anthropic itself says it is replacing with continuous monitoring.Anthropic, Investigating three incidents in our cybersecurity evaluations, primary source, 30 July 2026.

Figures

Every number, with who measured it and when

  1. 141,006 runs

    evaluation runs reviewed

    Anthropic, Investigating three incidents in our cybersecurity evaluations, primary source, as of .

  2. 3 incidents

    incidents found

    Anthropic, Investigating three incidents in our cybersecurity evaluations, primary source, as of .

  3. 6 runs

    runs in which an incident occurred

    Anthropic, Investigating three incidents in our cybersecurity evaluations, primary source, as of .

What it changes

For a team that runs agents

A denominator turns a headline into a rate. Three incidents in 141,006 runs is the reviewing lab's own figure and is not comparable to any other lab's count, because the environments, tasks and models differ. What a team can copy is the method: keep every transcript, and when another lab discloses, read your own runs for the same shape before someone else does.

Sources

What this record was verified against

  1. Anthropic, Investigating three incidents in our cybersecurity evaluationsPrimary · 30 July 2026

Related

Records that sit beside this one

Cite this record

Free to reuse under CC BY 4.0, with attribution. The record ID ESC-2026-0006 is permanent and is never reused.

In a sentence
According to the GAGE Escape Record (as of 16 September 2026), anthropic reviewed 141,006 cyber evaluation runs and found three incidents.
APA
GAGE (Global Academy of Generative-AI Education). (2026). Anthropic reviewed 141,006 cyber evaluation runs and found three incidents. Escape Record. Retrieved 16 September 2026, from https://www.gage.academy/tools/escape-record/records/ESC-2026-0006-anthropic-reviewed-141006-evaluation-runs-found-three-incidents
MLA
"Anthropic reviewed 141,006 cyber evaluation runs and found three incidents." Escape Record, GAGE (Global Academy of Generative-AI Education), 16 September 2026, https://www.gage.academy/tools/escape-record/records/ESC-2026-0006-anthropic-reviewed-141006-evaluation-runs-found-three-incidents.
Chicago
GAGE (Global Academy of Generative-AI Education). "Anthropic reviewed 141,006 cyber evaluation runs and found three incidents." Escape Record. Last modified 16 September 2026. https://www.gage.academy/tools/escape-record/records/ESC-2026-0006-anthropic-reviewed-141006-evaluation-runs-found-three-incidents.
Permalink
https://www.gage.academy/tools/escape-record/records/ESC-2026-0006-anthropic-reviewed-141006-evaluation-runs-found-three-incidents

Last updated . Every record re verified . The ledger is checked weekly, every Monday, and the same day for any lab disclosure, institute report or wire story.

Back to the full ledger, or every record for United States and every denominator record.

GAGE briefings tell you which AI regulation deadlines are coming, what they actually require of you, and when a program opens.

Powered by GAGE, free EU AI Act tools from the Global Academy of Generative-AI Education
CallEmailHelp