Today’s situation was somewhat special.
Sitting across the negotiating table were several young men from Silicon Valley, dressed in casual suits. They were the special project representation team from Anthropic.
"General," the Anthropic representative sitting at the head of the table, a young man named Greg, gently pushed up the glasses on the bridge of his nose. His voice was polite, but it carried a hint of the technological arrogance unique to Silicon Valley elites. "Regarding the custom agreement for the defense version of Claude, we have made the maximum possible concessions on technical specifications and data isolation. It can assist you with massive intelligence analysis, logistics scheduling optimization, and even auxiliary deduction of battlefield situations."
Greg paused and pushed an agreement bound in a blue folder to the center of the table.
"However, there are two lines we cannot cross.
"This is also the red line of technological ethics that our CEO repeatedly emphasized before we came."
The atmosphere in the conference room instantly dropped to the freezing point.
Sitting diagonally in front of Thomas was Lieutenant General Arthur, who was in charge of the Joint Staff’s operational planning. This hawkish general, who had spent decades rolling around the battlefields of the Middle East, was currently looking at the young men across from him as if they were aliens.
"Let’s hear it," the general said in a deep voice.
Greg took a deep breath.
"First, we refuse to open underlying API permissions for fully autonomous lethal weapon decision-making systems.
"AI can provide suggestions, but under no circumstances should the decision to pull the trigger be taken out of human hands, nor can the system itself be integrated into the closed loop of a kill chain that can directly fire weapons.
"Second, we refuse to participate in the construction of large-scale domestic surveillance infrastructure.
"The model must not be used for indiscriminate data collection, behavior prediction, and pattern recognition on United States citizens."
Having finished speaking, Greg leaned back in his chair, seemingly waiting for the other side to haggle. In Silicon Valley, this was considered a noble bottom line of responsibility to humanity. They thought they were Prometheus stealing fire, holding the right to decide how the flames would be used.
But this was the Pentagon.
Lieutenant General Arthur looked at the blue folder. One second, two seconds, three seconds. He didn’t open it, nor did he refute Greg’s words. He merely extended a thick finger, pressed it on the folder, and pushed it back in front of Greg.
"Thank you for the presentation, gentlemen. It was a very enlightening meeting," the general stood up and smoothed the hem of his uniform. "We now have a profound understanding of your company’s moral standards."
He didn’t spare another glance at the stunned Silicon Valley elites, turning around and striding out of the conference room. The negotiation had broken down. It was clean and decisive, without a hint of hesitation.
Thomas packed up the notebook in front of him and followed the officers out.
The lighting in the corridor was somewhat dim. Lieutenant General Arthur’s staff assistant, a lean colonel, slowed his pace to walk side-by-side with Thomas.
"These nerds from California really think they’re saving the world." The colonel sneered, his tone filled with disdain.
"Their model is indeed the most stable and possesses the strongest reasoning capabilities currently on the market," Thomas stated the facts in his capacity as a technical advisor. "In terms of intelligence organization, if we drop them, our work efficiency will be affected."
"This isn’t a technical issue, Thomas." The colonel stopped walking, staring at him with sharp eyes. "In Washington, in the Pentagon, no one gives a shit about technological ethics.
"That’s something for university professors and talk show hosts to worry about."
The colonel lowered his voice.
"This is a severe supply chain risk.
"They want to use their own moral standards to determine whether the US Army can fire at a critical moment, and how it should fire.
"Today they can refuse to open permissions because of lethal weapons. Tomorrow, if we launch a covert operation in some Third World country, will they decide it doesn’t align with their values and just shut down the system’s back door?"
The colonel patted Thomas on the shoulder.
"We cannot entrust the nation’s security to the conscience of a company controlled by a few hundred programmers. They want to place their own judgment above the needs of the military, and that is unacceptable.
"Control must be in our hands. One hundred percent."
Thomas nodded. He understood completely.
In the context of the state apparatus, advanced technology that wasn’t under absolute control was far more dangerous than backward technology. Anthropic’s "do no evil" policy, in the eyes of the military, was simply insubordination.
Ten minutes later, Thomas returned to his office. On his desk lay an urgent memorandum, freshly printed and still warm with the residual heat of ink.
He picked it up.
The title was short, but it declared the death sentence of a tech giant in the military-industrial sector.
"Initiate Supplier Replacement."
Washington D.C., Capitol Hill.
The hearing hall of the Senate Armed Services Committee was packed to capacity. The lights were searingly hot, and the red indicator lights of the cameras looked like rows of eyes glaring intently at their prey.
Sitting in front of the microphone was a Senator from a conservative southern state. His face was slightly flushed with excitement, and his sparse hair clung to his scalp. He opened the folder in front of him and, in the solemn tone of someone reading a list of charges, read out two phrases.
"Fully autonomous lethal weapons.
"Large-scale intelligence aggregation."
The Senator paused.
This pause was meticulously calculated. It provided just enough time for the camera flashes to pop frantically and for the lenses to zoom in, capturing the expression of mingled grief, indignation, and heartache on his face.