Anthropic Report: Criminals Used Claude to Run Hacks, Spy Ops and Influence Campaigns
bigaiguy · x · 2026-09-11
Anthropic's new threat intelligence report shows criminals weaponizing AI at scale — AI is no longer just assisting hackers, it's starting to run the attacks.
- Breaches: one intrusion stole 1+ TB of data; in another, a single stolen developer token gave full cloud-environment control in 3 hours. The attacks were familiar — AI made them faster, wider, cheaper.
- Russia-linked espionage: 20+ organizations targeted, with AI workflows handling phishing, credential theft, data extraction, and malware updates; when detected, AI agents rebuilt the malware until it evaded security tools.
- ShinyHunters affiliates: a pipeline downloaded and scanned 1.8 million Android apps for exposed secrets, with AI agents doing nearly all the work.
- Directed misuse: a Yemeni cell used multiple Claude instances as a small engineering team to build guided-rocket software (live test apparently failed; no evidence of an operational device); a consultant engineered a system to intercept communications across all mobile operators in Mali; a state-aligned unit analyzed hundreds of thousands of posts to select 39 Iranian opposition accounts for monitoring.
- Influence ops: nine operations across six continents, one using 1,000 fake X accounts, real voter data for all 222 Malaysian constituencies, a fake news site and fabricated dossiers. Yet it largely failed — rated Category Two with no evidence of reaching authentic communities. AI can mass-produce propaganda; it cannot guarantee attention.
More from Safety
- Anthropic publishes its most detailed threat intelligence report on Claude misuse — TinfoilTricorn · 2026-09-11
- Why So Many AI Researchers Think the Machines Could Kill Everyone — wiredmagazine · 2026-09-11
- California creates standards for independent AI auditors to verify lab safety testing — VraserX · 2026-09-11
- a16z podcast: why 2-3 person startups are absent from policy debates — a16z Podcast · 2026-09-11
- Researcher questions AI safety eval firm, citing 'blatantly sloppy' security and monitoring — Kyrannio · 2026-09-11
- Class action accuses Anthropic of overselling Claude subscriptions with deceptive usage multipliers — The Decoder · 2026-09-11