Skip to content
NewA calmer, faster ZeroHour

Agents That Edit Documents: Measuring Agentic PDF Forgery Against a Non-Agentic Control

arXiv cs.CRby Simiao Ren ()1 min read2 · 1 read
AI summary · glm-5.3Summaries by Vani AI

AI agents can forge financial documents with high success rates and low costs, though the threat is overestimated by a factor of two.

A study on AI agents forging documents shows agents can alter financial documents like invoices, achieving high forgery success rates and low costs, though the threat is overestimated by a factor of two.

Key facts
  • AI agents can forge financial documents by editing amounts, dates, and addresses
  • 1,419 out of 1,750 agent-forgery tests pass the verifier
  • 808 out of 1,250 tests survive stricter filters
  • deterministic scripts achieve 98% forgery success
  • agents achieve 124% forgery success
  • cost of verified forgery is 2.4 cents
Analyze this story with Vani AIImpact, who is affected and what defenders should do, in a new chat

From the original article

1 min read at arxiv.org

AI agents that carry a multi-step computer task through on their own became ordinary tools in the past year, and the same autonomy is available to anyone whose task is harmful. We ask what that means for a relying party -- an insurer, a lender, an auditor -- whose evidence is a filed PDF.

arXiv cs.CR

By Simiao Ren · arxiv.org

Read the full article at arXiv cs.CR

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.