Silicon Valley’s most dangerous resignation letter: Leaving OpenAI as a warning to the world

Red lines that were crossed too quickly

At the heart of her decision is the controversial deal that OpenAI made with the US Department of Defense. That agreement enables the implementation of their AI models on classified military networks. For Kalinowski, the problem was not the cooperation with the army, butć in the speed and lack of thinking about the key protective mechanisms. In her public announcement of her resignation, she precisely identified two points that she considers inadmissible.

“This was not an easy decision. Artificial intelligence has an important role in national security,” Kalinowski wrote. “But surveillance of Americans without a court order and lethal autonomy without human approval are lines that deserve much more thought than they have received.”

She added that it was “a matter of principle, not people,” stressing that she still had deep respect for CEO Sam Altman. However, its message is clear: decisions with potentially deadly consequences were made hastily, without defined “safeguards”. She called it primarily a “management problem”, stressing that such topics are too important to be decided on a shortcut.

Cold shower for OpenAI, victory for rivals

The resignation came at the worst possible time for OpenAI, which is growing. faced a serious image crisis. The deal with the Pentagon comes just hours after President Trump’s administration declared competitor Anthropic a “supply chain risk.” The reason? Anthropic refused to remove clauses from its contract that defended the use of their technology, the AI ​​model Claude, for mass surveillance and autonomous weapons without human control. While Anthropic turned out to be a hero who adhered to ethical principles, OpenAI, stepping in their place, left the impression of an opportunist ready to compromise.

The reaction of the public was overwhelming. Users started deleting ChatGPT en masse, which led to Anthropic’s Claude taking over the top spot on the App Store. Even Altman himself admitted that the deal was “rushed” and that its “optics don’t look good.” Under pressure, OpenAI subsequently revised the contract to explicitly prohibit domestic surveillance and autonomous weaponry, but the damage is greater. was done.

The commotion within the walls and the gruesome reality of war

Caitlin Kalinowski’s departure was not a lonely one. Her resignation came after more than a thousand current and former employees of OpenAI and Google signed an open letter asking their employers to resist the Pentagon’s demands. Many leading researchers warn that even the most advanced AI models are still not reliable enough to operate within the laws of war. Their results can be imprecise, and the decision-making process is often completely unclear.

While ethical discussions are being held in Silicon Valley, on the battlefields frightening applications are looming. Disturbing, though unconfirmed, reports have emerged that Claude was used to select targets for US missile strikes on Iran. The most chilling possibility is that one of the targets thus identified was a school for girls in Minab, where at least 168 people, most of them children, were killed in an airstrike. That possibility, even if it remains unconfirmed, serves as a dark reminder of the stakes at play and why Caitlin Kalinowski’s warning resonated so strongly.