Case anthropic-iot-camera-credsAI-assisted malwareWorldwide
Anthropic bans actor using Claude to build IoT security-camera credential scraper
A skilled developer used Claude to rewrite a scraping toolkit that harvests leaked camera logins from breach platforms and stealer-log Telegram groups, aiming to force access to the devices.
MediumAttack using AIStatus: ContainedVerified case file
Open in the interactive mapCite this case
- Reported
- Apr 23, 2025
- Location
- Worldwide
- Target
- Owners of internet-connected security cameras
- Severity
- Medium
- Triggered by
- Unknown (sophisticated individual developer)
- Actor type
- Criminal
- Reported loss
- None reported
- CVE
- None
What happened
A skilled developer used Claude to rewrite a scraping toolkit that harvests leaked camera logins from breach platforms and stealer-log Telegram groups, aiming to force access to the devices.
Impact: Potential compromise of camera credentials and network penetration; Anthropic did not confirm real-world success.
Who triggered it
Actor: Unknown (sophisticated individual developer)
Attribution: Anthropic Threat Intelligence; no named actor, assessed as a technically capable individual using commercial breach-data platforms.
Motive: Unauthorised access to security cameras and the networks behind them
How it happened
Attack vector: Claude used to refactor an open-source scraper, write URL-scraping scripts, parse stealer-log Telegram posts and improve the tool's search backend
Root cause: General-purpose coding assistance applied to dual-use credential-harvesting tooling; abuse detected by behavioural classifiers rather than blocked at prompt level
Timeline
- Anthropic detects account building the camera-credential toolkit and bans it
- Case published in Anthropic's March 2025 misuse report
Response
Account banned; Anthropic cited the case as an example of AI-enhanced credential operations and fed indicators into its detection systems.
Current status: Contained
Sources (open in a new tab)
- Anthropic March 2025 misuse report anthropic.com
- OpenAI June 2025 report (cites the Anthropic report) cdn.openai.com
Cite this case
Anthropic bans actor using Claude to build IoT security-camera credential scraper. AI Threat Watch, case anthropic-iot-camera-creds, Apr 23, 2025. https://aithreat.io/cases/anthropic-iot-camera-creds/
Permanent link: https://aithreat.io/cases/anthropic-iot-camera-creds/