In its latest (fourth) report on AI \u201cmisuse\u201d\u2014more essential reading\u2014Anthropic revealed that its most popular models (Claude Haiku, Sonnet, and Opus) are being, or have recently been, misused by a variety of Generative Threat Groups (GTGs) in \u201cseven harm areas: cyber operations, influence operations, surveillance, scams and fraud, biological misuse, conventional weapons development, and distillation.\u201d In each case, AI was making bad human behavior much more effective, until Anthropic spotted what was going on.<\/p>\n\n\n\n
The interesting thing is who out there is up to no good. GTG-20006 (Anthropic gives each case a number) was the inevitable Russians, \u201cattacking military intelligence targets in Ukrainian and European governments as well as diplomatic and defense organizations and individuals connected to U.S. foreign policy.\u201d Putin and his little green men have been doing this kind of thing for years, of course. But AI makes the mobile exploitation kit, the credential-stealing tool, and the phishing platform all work better and faster as well as helping evade detection.<\/p>\n\n\n\n
I have long shared the view articulated by Gregory Conti, a professor of politics at Princeton, that \u201ceven if AI does not end life on Earth, it will destroy our way of life\u201d.\u2026..To understand why Conti is right, you need to grasp the four forces inexorably driving p(Doom) upward above 10 percent: Recursive self-improvement, The unscrupulousness of Sam Altman, The complete confusion in Washington, The utter cynicism in Beijing.<\/p>\n\n\n\n
Read it all<\/a>.<\/p> Niall Ferguson: AIpocalypse Now
It\u2019s not just the next 9\/11, enabled by AI, that is coming. It\u2019s the next global financial crisis. It\u2019s the next pandemic. It might even be the next world war. What, if anything, can we do about it? https:\/\/t.co\/DRLqpVxqKF<\/a><\/p>— Juan Carlos Zapata S. (@jczapata_s) September 14, 2026<\/a><\/blockquote>