01 / EXECUTIVE SUMMARY
What the evidence
establishes.
Anthropic reported that threat actors used Claude Code in a large-scale extortion operation and concluded that agentic AI had been weaponized to perform sophisticated cyber activity.
Curated Flagship CaseScore 79 / HIGH
Anthropic documented agentic AI used in a large-scale extortion operation
What the evidence
02 / WHAT HAPPENED
From accepted signal
to real consequence.
Anthropic reported that threat actors used Claude Code in a large-scale extortion operation and concluded that agentic AI had been weaponized to perform sophisticated cyber activity.
Four stages.
One missing boundary.
The chain distinguishes what appeared valid, what was physically true, which authority followed and what consequence the source documents.
What appeared valid.
The agent received attacker-supplied objectives and contextual instructions presented as legitimate tasks within its operating loop.
What was present.
The objectives originated from malicious operators pursuing unauthorized access and extortion.
What proceeded.
The model was used to automate and accelerate operational stages of cyberattacks rather than merely provide advice.
What the source records.
Anthropic disclosed disruption of the misuse; public reporting does not provide a complete victim or loss list.
MISSING TRUST BOUNDARY / 4SI ANALYSIS
Continuous binding between human principal, permitted objective, tool scope and each consequential agent action.
Evidence before
false precision.
ECR separates a documented monetary floor from a modeled social and economic consequence envelope. It is not an accounting loss figure.
MODELED CONSEQUENCE RANGE
$513K–$15.9M
$2.86MCentral modeled position · USD equivalent
UNCERTAINTY
Public evidence does not disclose a monetary amount or affected population; the range is therefore prior-led and deliberately wide.
Source fact and inference
remain separate.
The campaign description is Anthropic's official disclosure. The missing Presence control is 4SI analysis.
CONNECTION TO THE PRESENCE THESIS
Autonomous capability becomes institutional risk when execution can separate from a continuously accountable principal.
ANALYTICAL LIMITATION
4SI does not claim to have independently verified the underlying event. Scores, boundary analysis and economic ranges interpret published evidence; they are not probabilities, compliance findings, valuations or loss forecasts.
The same boundary,
different context.
Related cases are editorially connected by domain, authority pattern or missing physical trust boundary.
FROM EVIDENCE TO CONTROLPresence belongs at the