Hackers Impersonate ChatGPT, Claude, and Perplexity AI Bots to Hunt for Passwords and API Keys
The campaign ran from late September to late October, impersonating 32 AI bot identities to scan for exposed secrets across multiple industries.
Attackers impersonated trusted AI bots including ChatGPT, Claude, and Perplexity to scan enterprise websites for exposed passwords, API keys, and cloud credentials. The single actor used over 300 cloud addresses and rotated through bot identities to request paths for sensitive files like .env files and SSH keys. Cequence tracked the activity and noted it reached 48% of protected organizations on peak days, but found no evidence of widespread credential theft.
Unconfirmed: The findings do not establish widespread credential theft.
- Scanner rotated evenly through 32 bot identities from OpenAI, Anthropic, and Perplexity.
- Used bursts of 50-190 requests in eight seconds across 300+ cloud addresses.
- Targeted financial services, retail, telecommunications, energy, and software firms.
- Requested random nonexistent paths to identify responsive websites.
- Nearly all requests were blocked or failed at target organizations.
Impact: exposed passwords, API keys, and cloud credentials via impersonated AI agents
- 1.Verify AI bot identities via signed requests or published network ranges
- 2.Remove exposed secrets from web-accessible locations
- 3.Evaluate access requests by behavioral patterns rather than labels
- AI models
- ChatGPTClaudePerplexity
From the original article
2 min read at gbhackers.com
Attackers are impersonating trusted AI bots to scan enterprise websites for exposed passwords, API keys, and cloud credentials. The company tracked a single actor rotating through approximately 32 bot identities, including ChatGPT, Claude, and Perplexity, to probe organizations across multiple industries. On one October day, the campaign reached 48% of Cequence-protected organizations that received any […]
By Mayura Kathir · gbhackers.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.