Anthropic Gets Subsea Cable Security Wrong, Coalition Tells FCC
The tech company’s FCC filing shows a misunderstanding of subsea cable vulnerabilities and regulations, the group says.
WASHINGTON, May 29, 2026 — Did someone at Anthropic forget to consult its AI assistant Claude?
In a May 27 letter to the FCC, a group called the International Connectivity Coalition said that Anthropic’s public comments on subsea cable security contained various technical errors.
Submarine cable connectivity “bears no resemblance to the open, internet-facing exposure Anthropic implies,” the coalition said, adding that Anthropic’s hacking concerns were “unsupported by any evidence in the record.” Anthropic is led by CEO Dario Amodei.
The ICC disagreed with specific issues in Anthropic’s filing with the FCC, such as the company’s claims the cables could be throttled, tapped, or otherwise exploited by foreign adversaries.
“Anthropic asserts that owners or operators of submarine line terminal equipment could engage in service manipulation, including prioritization, throttling, or routing changes, to selectively degrade AI workloads,” the ICC said. “This assertion is incorrect as a technical and operational matter.”
The coalition, which includes Amazon, Meta, Microsoft, and Verizon, also warned the FCC against accepting Anthropic’s regulatory suggestions, which the group said exceeded the FCC’s statutory authority.
In a January 22 letter, Anthropic told the FCC that AI systems operated by foreign adversaries could compromise encrypted data and identify unknown infrastructure “backdoors.”
But the coalition argued that Anthropic’s concerns of foreign ownership “fail to account for limits already adopted by the [FCC],” making new rules “duplicative and unnecessary.”
The FCC has sought to update its submarine cable licensing rules in response to concerns of foreign ownership. Anthropic and other companies filed comments in response to the agency’s notice of proposed rulemaking in August 2025.
Anthropic, which makes the Claude family of large language models, is currently the most valuable startup company in Silicon Valley at $965 billion, according to a recent CNBC report. Anthropic’s Mythos model made headlines in April for its ability to find previously undetectable software vulnerabilities.
