中文翻译

摘要: Global

General Business

Technology

Anthropic AI Model Identifies Vulnerabilities in Classified U.S. Government Systems During Testing

June 24, 2026

Anthropic

‘s advanced artificial int...

正文

General Business

Technology

Anthropic AI Model Identifies Vulnerabilities in Classified U.S. Government Systems During Testing

June 24, 2026

Anthropic

‘s advanced artificial intelligence model,

identified vulnerabilities in highly sensitive U.S. government computer systems during a recent testing exercise conducted with U.S. intelligence agencies, according to a report by The Associated Press.

A U.S. official familiar with the effort told AP that Anthropic’s Mythos model uncovered certain weaknesses within hours of testing. The official noted that finding vulnerabilities did not necessarily mean the AI could exploit them during that timeframe.

The testing was conducted through

Anthropic’s Project Glasswing

initiative, which brings together technology companies and government partners to evaluate and mitigate potential cybersecurity risks posed by increasingly capable AI systems. The program aims to help protect critical software infrastructure from threats that could affect public safety, national security and the broader economy.

The effort was referenced publicly earlier this month by

Sen. Mark Warner,

D-Va., during a Senate Committee on Banking, Housing and Urban Affairs hearing. Warner said an advanced AI tool was able to penetrate nearly all classified systems tested “not in weeks but in hours,” citing information provided by National Security Agency and U.S. Cyber Command leadership.

The testing comes amid growing scrutiny of advanced AI models and their potential cybersecurity implications. While Anthropic has worked with U.S. agencies to assess vulnerabilities, the company has also expressed concerns about military applications of artificial intelligence, contributing to tensions with the Trump administration.

Earlier this month, the administration directed Anthropic to restrict access to its latest AI models, Fable 5 and Mythos 5, for foreign nationals. The move followed President Donald Trump’s executive order es

来源: Brave/citybiz.co

采集时间: 2026-06-24 20:16:51