The "duel" in the AI field is evolving from a technological competition to an open public battle. According to media reports citing internal sources, Dario Amodei, co-founder and CEO of Anthropic, recently launched a fierce attack on OpenAI and its CEO Sam Altman in an internal memo to employees, accusing them of making "outright lies" regarding their statements on handling U.S. Department of Defense collaborations.

This conflict originated from the different attitudes of the two companies towards military contracts. Previously, Anthropic failed to reach a new agreement with the military because it insisted that the Department of Defense commit not to use AI for domestic mass surveillance or autonomous weapons. Subsequently, the Department of Defense (also known as the "War Department" in certain political contexts) signed a contract with OpenAI. Altman claimed that OpenAI's new contract also included the same safety red lines that Anthropic had insisted upon.

Amodei described OpenAI's approach in his letter as "a safety game." He wrote, "OpenAI accepted the agreement to appease its employees, while we did it to truly prevent abuse." He accused Altman of disguising himself as a "peacebuilder" and "agreement maker," but actually spreading misleading information, calling his statements "pure lies."

The core of Amodei's anger was the definition of "legitimate use" in the contract. Although OpenAI stated in its official statement that the contract clearly excluded illegal surveillance, Amodei believed this statement was "brainwashing" and "manipulating public opinion," as legal standards may change with policy shifts, and what is "legal" does not necessarily equate to "safe."