DoubleChecks

What was claimed

The AI threat everyone feared is already happening: AI used to build mass surveillance of 25M SIM cards, develop missile guidance systems (including hypersonic) in Yemen, fully automate hacking/phishing/malware iteration, run propaganda at scale, and conduct dual-use bio research

Our verdict

Accurate

A Mali consultant used Claude to help build surveillance covering roughly 25 million SIM cards for Mali's state intelligence service. This is documented in Anthropic's September 2026 threat report. (Only 1 of 3 AI systems responded.)

Assessment28 sourcesChecked Sep 12, 2026

Why this verdict

  • Conduct dual-use bio research

    Verified
    One assessment reached this conclusion

  • Run propaganda at scale

    Verified
    One assessment reached this conclusion

  • Fully automate hacking/phishing/malware iteration

    Verified
    One assessment reached this conclusion

  • AI used to build mass surveillance of 25M SIM cards

    Verified
    One assessment reached this conclusion

Evidence reviewed

28 sources

Claude source is combined here.

yellow.comfonearena.comyahoo.comen.cryptonomist.chthe-decoder.comal-monitor.comdigitaljournal.comaicybr.commagnifypost.comglobalsecurity.org

Individual assessments

PerplexityDid not respond

Assessment summary

  • The statement accurately summarizes findings from Anthropic's September 2026 threat intelligence report.
  • All five major claims are supported by documented evidence from the report, though the phrasing "already happening" appropriately reflects that these are real incidents that have occurred.

Claims reviewed

Conduct dual-use bio research

Verified
Researchers in unsupported regions used proxy networks to route dual-use scientific queries through other systems, including research involving pathogens, immune evasion and other high-risk biological topics.

Run propaganda at scale

Verified
Over just two weeks in April 2026, Anthropic found more than 4,700 different Claude-powered AI characters chatting with at least 25,000 real users, sending about 2.36 million messages in total, which represents large-scale automated propaganda/fraud operations.

Fully automate hacking/phishing/malware iteration

Verified
Chinese-speaking operators used Claude for offensive cyber work, including vulnerability research, with one continuously running workflow generating more than a dozen possible zero-day findings in a single month, while parallel systems handled reconnaissance, malware development and intelligence collection. Additionally, research evaluated large language models' capability to conduct fully automated spear phishing campaigns, with researchers developing and testing an AI-powered tool to automate spear phishing using AI agents based on GPT-4o and Claude 3.5 Sonnet.
ChatGPTDid not respond

Check your own claim

Paste any statement, headline, or AI answer. Independent assessments check it and compare sources.

Share this result