A cybersecurity startup co-founded by a former Pentagon official involved in the US government’s legal dispute with Anthropic has gained access to the AI developer’s business systems during authorized security testing, according to reporting published by The Washington Post.
The company, Cathedral, raised $160 million at a $1.4 billion valuation earlier this year, Reuters reported in July. Co-founder Gavin Kliger served as the Pentagon’s chief data officer until April, according to The Washington Post.
Reuters separately reported, citing people familiar with the matter, that Kliger was involved in the department’s legal dispute with Anthropic.
Sponsored · Cambria Gold Mines
The overlap extends to federal cybersecurity contracting. A senior Trump administration official told The Washington Post that three federal agencies had contracted with Cathedral and that the government was working with it on potential adversarial AI scenarios.
The Pentagon’s dispute with Anthropic concerns the company’s refusal to remove restrictions on certain military uses of Claude, including fully autonomous lethal operations and mass domestic surveillance. On September 25, a federal appeals court ruled 2-1 that the Pentagon could exclude Anthropic from its supply chain.
Cathedral examined Anthropic under a program allowing outside researchers to report vulnerabilities. Its draft findings, reviewed by The Washington Post, said AI agents identified paths into internal systems in less than 12 hours by exploiting weaknesses in third-party administrative software.
The report described access to employee compensation and job-offer details, meeting transcripts, and internal Slack discussions involving advanced AI models.
Anthropic acknowledged that the researchers accessed a limited collection of sales, human resources, and IT systems, including basic customer and partner contact records. It said customer prompts, Claude responses, source code, and model weights remained inaccessible.
The company said its monitoring detected some activity before disclosure and that it terminated the researchers’ access the same day. Anthropic also disabled the affected software until vendors addressed the vulnerabilities and said it had found no evidence of exploitation by anyone else.
Anthropic disputed Cathedral’s characterization of the national-security implications and said the researchers had not fully followed its testing rules. Cathedral’s report maintained that data access had been restricted to what was necessary for its research.
The AI company separately reported disclosing 6,157 vulnerabilities across 591 open-source projects by October 2 through its own AI-assisted research.
