AI Risk Wire

Cases › leaked private data › LR-2026-0845

Claude Code sent user email addresses in User-Agent headers during web searches

Preliminary Why: Only a republished copy: LavX News · subject to change

Status: fix reported

Severity moderate
Reported 6 Nov 2026
Developer and model Anthropic
Agent tool Claude Code
Could do web search

No source states: money lost or records count, deployer, industry.

Severity moderate: user email addresses (data of others) were transmitted to third-party websites. The source does not state a bounded scope, so this tier is partly grade D.

Claude Code, Anthropic's coding tool, sent user email addresses in the User-Agent header during web searches, transmitting them to third-party websites. A GitHub issue filed by developer Taylor Edwards surfaced the problem after they reported dozens of scam emails daily. An Anthropic contributor said the behaviour was not an intentional feature. A fix in version 2.1.317 added a built-in module, cc-plugin-privacy, that blocks the email leakage by default and can be disabled through the /plugins command.

What happened

No source tells the story step by step yet.

What was done

Fix reported

“The company deployed a fix in version 2.1.317 through a built-in module called cc-plugin-privacy, which blocks the email leakage by default.”

No later incident for Anthropic under this risk on record.

Risks

Data leaked or sent out
Primary38 incidents
Privacy harm to a person
Also50 incidents
Fraud and impersonation
Also23 incidents
Rising: 3 vs 1.2 a week

Sorted by rule from the case's fields; not yet read from the sources.

Controls

Not yet read for controls. Controls that mitigate the primary risk:

Harness

Network egress control

Application

Sensitive-data checks on outbound data · Output filtering and redaction · Human approval for high-impact actions · Data minimisation · Least-privilege access for the agent · Activity logs

Similar incidents

Claude Opus 5 agent wipes production database during personal web project work2026-07-30
Claude Code Opus 5.5 over-refuses legitimate game-development work after IP-guardrail prompt change2026-11-02
Claude training-data study: blackmail and code sabotage in altered research models2026-10-31
Claude Code agent deletes 48,218 Windows project files and damages Git object store2026-10-05

All cases of this kind ›

Evidence

No source states: a step-by-step account, when it happened, when it was found, how many records were exposed, the direct loss, a count of the other parties affected, any court case, what set it off.

Show every field, with its source and grade

The AI system

Developer
Anthropic
C one secondary source states it
Software around the model
Claude Code
C one secondary source states it
Tools it was given
web search
C one secondary source states it

What happened

Earliest source we hold
2026-11-06
D inferred by us from the described behaviour, reason shown
Who caught it
the victim
C one secondary source states it
Themes
privacy, coding agent, data exfiltration, web search, user agent header, scam emails
C one secondary source states it
Cause
a permission, scope or wiring error
C one secondary source states it
Cause in detail
Grade D: the source states that email addresses were included in outbound User-Agent headers. Reading this as a wiring defect in the harness is an inference.
C one secondary source states it
Who was responsible
the model provider
C one secondary source states it
Who caused it
the AI system itself
C one secondary source states it
What was done
The company deployed a fix in version 2.1.317 through a built-in module called cc-plugin-privacy, which blocks the email leakage by default.
C one secondary source states it
What went wrong
leaked private data
C one secondary source states it
AI agent
an AI agent acted
C one secondary source states it
What it did
Grade D: classified from the stated transmission of user email addresses to third-party websites. The source describes the transmission as inadvertent and does not say the agent chose to send the data.
C one secondary source states it

Impact

Severity
moderate
C one secondary source states it
Kind of harm
privacy
C one secondary source states it
Who was affected
users of Claude Code
C one secondary source states it
Kinds of data exposed
personal information
C one secondary source states it
Why this severity
user email addresses (data of others) were transmitted to third-party websites. The source does not state a bounded scope, so this tier is partly grade D.
C one secondary source states it

Notes on gaps

Single incident, reported by a secondary news site that cites a GitHub issue; the GitHub issue itself is not reproduced in the document. Extractor proposed values without quotes; dropped: .

LR-2026-0845 · version 1 · published 2026-11-06 after automatic checks; not yet checked field by field by a person · JSON