Topic

AI Misuse Threat Intelligence

Reporting by AI labs on detected and disrupted attempts to misuse their models for cyber operations, surveillance, influence, weapons development and model distillation.

Current clusters

build3 publishers

OpenAI gated an 88-hour, 10,000-agent proof search on a 17-hour Lean check

OpenAI's write-up gives the token counts, the agent count and the verification time for its Navier-Stokes result. The verification time is the number that decides whether the method transfers to anyone else's workload.

Publishers:dev.tosimonwillison.nettheneuron.ai

Perspective Coverage

3 publishers

Builder: Builder 38%
Operator: Operator 45%
Investor: Investor 17%

Reality

Evidence62
Adoption20
Hype gap+22
Incentives78
Confidence66

invest2 publishers

An Iran-linked actor built US Navy targeting handbooks from public data with Claude's help

Anthropic's threat report says the operation pulled transponder identifiers, satellite-imagery scripts and personnel names off captions on public military photographs, and the company banned the account it traced the work to.

Publishers:indianexpress.comqz.com

Reality

Evidence38
Adoption44
Hype gap+18
Incentives74
Confidence50