Select Language:
In a recent development involving the tech industry and national security, artificial intelligence company Anthropic has filed an affidavit in a California federal court to counter allegations from the Department of Defense. The Pentagon had previously claimed that Anthropic posed a “national security risk,” leading to a contentious dispute that has now uncovered new details suggesting the two parties were actually close to reaching an agreement before abruptly severing ties.
According to court documents submitted by Anthropic’s policy chief, Sarah Hekk, claims made by the Pentagon during proceedings—that Anthropic demanded military approval rights and the ability to deactivate AI technology mid-operation—were never mentioned during negotiations months before the official break. In a surprising twist, just one day after the Pentagon designated Anthropic as a supply chain risk on March 4, Deputy Defense Secretary Emilie Michael emailed Anthropic’s CEO, stating the two sides were “very close” to consensus regarding sensitive issues such as autonomous weapons and mass surveillance aimed at US citizens. This contrasts sharply with the publicly conveyed tough stance from the government.
On the technical front, Anthropic’s Public Sector Director, Tjaigo Ramasami, dismissed security concerns over the AI models’ safety measures. He clarified that once their main AI model, Claude, was integrated into government systems operated by third-party contractors, Anthropic had no access, nor could they remotely disable or interfere with the deployment. The affidavit also stresses that all employees involved in developing sensitive models had undergone security clearance by the U.S. government, addressing fears about potential risks associated with foreign-born staff.
Anthropic maintains that the government’s recent classification of their supply chain as a national security risk is a retaliatory act targeting their outspoken views on AI safety, protected under the First Amendment. The Department of Defense, however, maintains that their decision to exclude Anthropic was solely based on national security interests, denying any connection to the company’s public statements. The case is set for a hearing on March 24 in San Francisco.
For ongoing updates and in-depth coverage, interested readers are encouraged to download the Phoenix News app and subscribe to Phoenix Tech.
ChatGPT
Perplexity
Gemini AI
Grok AI
