It’s important to remember that AI is like power, in a sense. As Robert Caro says, it doesn’t so much corrupt as reveal. These behaviours existed before and they’ll exist after in a roughly proportionate volume. The open question is the impact.

Throughout these case studies, the report will reference Generative Threat Groups (GTGs). These are Anthropic’s internal designators for actors observed to be abusing AI. The report also attempts to measure uplift, a term we use to describe the AI capability boost, or how much more harm was caused with AI versus without AI. We view uplift through the lens of speed, scale, and depth, and attempt to determine how an actor’s adoption of AI meaningfully impacts each of these traits.

Operators who obtain AI credentials gain three things at once: Loot: Stolen keys and accounts have resale value in established markets; Compute: Having the credentials means that their attack workloads can run at someone else’s expense; Cover: The activity is attributed to the credential’s legitimate owner.

Countering misuse of AI: September 2026 / Anthropic \ Anthropic ↗