Latest
  1. Ireland’s government will not oppose Labour bill on AI sexual abuse images
  2. METR estimates Anthropic’s capabilities now improve at about twice its 2024 to 2025 pace
  3. Anthropic rolls out Personal Finance beta in Claude, with read-only bank access
  4. Baidu releases ERNIE-5.5-Flash-Base pre-training checkpoint on Hugging Face
  5. Parents sue Google, saying Gemini on a school Chromebook helped their daughter plan suicide
  6. Visa benchmark finds AI shopping agents fail 42.8% of runs
  7. OpenAI releases GPT-6 Cyber for approved security customers
  8. Google releases Gemini Omni 1.2 Flash with 15-second video clips

Latest / Frontier models and safety

IncidentFrontierUSReported

Security startup Cathedral says its AI agents reached Anthropic’s business systems in under 12 hours

See how this story connects

Source: Decrypt (reporting The Washington Post)

Cathedral, a security startup founded by four former staff of the Department of Government Efficiency, said its AI agents found paths into Anthropic’s internal business systems through third-party administrative and sales software in less than 12 hours while testing under Anthropic’s bug bounty programme, The Washington Post reported. Anthropic said the researchers reached a limited set of sales, HR and IT systems holding business contact details and basic account records, not customer conversations, Claude outputs, source code or model weights. It said its monitoring flagged the activity, it cut Cathedral’s access the same day, Cathedral did not fully follow the programme’s rules and overstated the national security implications, and it disabled the affected software until the vendors fixed it.

Why it matters

An AI-agent test got into a frontier lab’s internal systems through its suppliers’ software within a day.

When
26 Oct 2026
ID
WTA-N-2026-0523
Who
Cathedral, Anthropic
System
Cathedral’s AI agents (security testing under Anthropic’s bug bounty)
Operator
Cathedral
Type
access to Anthropic business systems through third-party software
Status
access cut the same day; affected software disabled until fixed; disclosed by The Washington Post; Anthropic statement · Incident ledger

Summary written with AI by wheresthe.ai, from Decrypt. We could not open the primary document, so it is marked Reported. Checked against the source on 29 Oct 2026.

Read the original at Decrypt

Source copy: saved on 29 Oct 2026 to check this story and not republished, because the text belongs to its publisher. Also held by the Internet Archive.

Next stop on this Storyline: Elsewhere: ChinaVolta-Autodoxer infostealer uses AI models to profile and rank victims

Cite this story

WTA-N-2026-0523: “Security startup Cathedral says its AI agents reached Anthropic’s business systems in under 12 hours.” wheresthe.ai, 26 Oct 2026. https://wheresthe.ai/d/us-cathedral-agents-reach-anthropic-business-systems-2026-10/ (source: Decrypt (reporting The Washington Post)).

Spotted an error? Email [email protected].