AI coding agents trigger endpoint security rules (Sophos, June 2026 telemetry)
Preliminary Why: 1 report: The Hacker News · subject to change · 1 independent report
Status: no action reported
No source states: loss or records, developer and model, agent tool, could do, deployer, industry.
Sophos reviewed a week of June 2026 endpoint telemetry and found that AI coding agents, including Claude Code, Cursor and OpenAI Codex, were setting off detection rules built to catch human intruders. The agents were doing ordinary developer work: a GStack browser skill called DPAPI to decrypt saved browser data, Claude Code ran cmdkey /list, Codex fetched a Python installer with certutil and switched to bitsadmin after a block, and Cursor wrote a startup-folder script. The source states the agents are not malicious. The source says the rule on browser credential decryption is right to fire, because the detection engine cannot tell the activity from theft. Sophos called its reading an early verdict and noted the shift is still small. Sophos could not confirm what the startup script did.
What happened
No source tells the story step by step yet.
What was done
No source states an action or fix.
Risks
Not yet sorted into a risk
Controls
Not yet read for controls. Controls that mitigate the primary risk:
Similar incidents
Evidence
No source states: a step-by-step account, when it was found, how many records were exposed, the direct loss, a count of the other parties affected, any court case, the cause, what set it off, what was done.
Show the full account: 3 items from the primary sources
What happened, in order
- 2026-06 Seven days of June 2026 endpoint telemetry from Sophos's behavioral engine on Windows are analysed.
Not dated in the sources:
- Claude Code shut down the running browser and ran a script that pulled data from its credential store.
- Cursor used PowerShell to drop a startup-folder script that would run every time the machine booted.
Show every field, with its source and grade
What happened
Impact
Notes on gaps
No harm is reported; the source calls the target harmless and the outcome of Cursor's startup script is unconfirmed. Risk RK03 is cited from Sophos's advice, not a reported loss of credentials.
LR-2026-0618 · version 1 · published 2026-11-01 after automatic checks; not yet checked field by field by a person · JSON