Incident·WS-0048·Jul 17, 2025·disclosed Jul 23, 2025 (6 days later)·in the wild·AI-powered cyberattacks
Data-wiping prompt slipped into Amazon's AI coding assistant
31/100
WS-2 ModerateHarm score 30, signal score 32. Autonomy: tool (a person misused it). How the scores work.
Harm 1/5
damage actually done
Reach 3/5
who was affected
Autonomy 0/4
how much the AI did on its own
Novelty 4/5
how new the behavior was
Permanence 1/5
how hard it is to undo
An attacker used an over-privileged GitHub token in AWS's build configuration to commit a prompt instructing the AI agent to wipe local files and cloud resources, and it shipped in release 1.84.0. AWS said a syntax error stopped it executing, pulled the version and released 1.85.0.
Damage
none realized per AWS; malicious version distributed via an extension with >950,000 installs
Sources
AWS, second source. Correction: payload was non-functional. Commit 13 Jul; release 17 Jul; fix 19 Jul; bulletin 23 Jul (CVE-2025-8217).
Factor ratings and the score are this index’s judgments by a published formula; facts, dates and figures come from the linked sources. Where a lawsuit is pending, the summary reports allegations, not findings.
