Defense contractor employee's personal Claude agent ran on a secure Pentagon system for two months
Wired AI · rss · 2026-10-30
WIRED, citing sources, reports that an Accenture employee's personal Claude AI agent ran on a Department of Defense network for two months before it was identified and the vulnerability could be patched. The incident exposes a governance gap as AI agents proliferate: personally-configured agent tools can bypass approval processes to reach secure systems and keep running unsupervised for months before detection.
More from Safety
- Scanning 25,267 MCP servers found 73% vulnerable, 143.5K vulnerabilities — anacondainc · 2026-10-30
- DeepSeek's cyber model enters internal testing; observers back keeping it closed — Xianbao_QIAN · 2026-10-30
- IIT Madras Professor Argues India Needs an Independent Agency to Assess AI — ravi_iitm · 2026-10-30
- Reddit will now require an account to access, CEO confirms login wall — tech234a · 2026-10-30
- TriFHE progress stalls as existing FHE schemes already use dense ternary secrets — fractalmoto · 2026-10-30
- Anthropic delayed pre-training over a month for security — but it won't slow AI, researcher argues — thlarsen · 2026-10-30