Misuse and incidents

OpenAI disrupts AI-assisted cyber, scam and influence networks

OpenAI documented and banned real accounts using ChatGPT in malware development, phishing, organized scams, authoritarian surveillance proposals, and covert influence operations. The actors mostly added AI to established workflows; OpenAI found no novel offensive capability and reported that direct malicious requests were refused.

0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

CURRENT ASSESSMENT · REVISION 1
TOWARD DOOM32confidence 86/100

Why it moved the index

Observed threat actors integrating ChatGPT into real cybercrime, fraud, surveillance proposals, and influence workflows is direct misuse evidence, while refusals, account disruption, and the absence of novel tactics materially limit its magnitude.

AUDIT TRAIL

Assessment history

  1. R1
    Toward 32 · confidence 86

    Adds the distinct October 2025 quarterly threat report after full incident and mitigation review.

    14 Aug 2026
SHARE THE FINDINGS

Share this page

DoomBench social sharing card for OpenAI disrupts AI-assisted cyber, scam and influence networks.
  1. DoomBench assesses “OpenAI disrupts AI-assisted cyber, scam and influence networks” as evidence moving toward doom, with magnitude 32 and confidence 86 out of 100 in the misuse and incidents category.

  2. The DoomBench assessment of “OpenAI disrupts AI-assisted cyber, scam and influence networks” is based on reporting from OpenAI and records the editorial rationale, source quality, attribution, and revision history.

  3. DoomBench summarizes “OpenAI disrupts AI-assisted cyber, scam and influence networks” as follows: OpenAI documented and banned real accounts using ChatGPT in malware development, phishing, organized scams, authoritarian surveillance...

    https://www.doombench.com/news/openai-disrupts-ai-assisted-cyber-scam-and-influence-networks-2025-10-07