Please follow us on Gab, Minds, Telegram, Rumble,Gettr,Truth Social, Twitter
Over the weekend, news broke that OpenAI reached an agreement with the Department of War, resulting in the Pentagon gaining model access. Skynet incoming...
The Department of War is using OpenAI. Of course, they’re using AI, and we know that their industry-specific tools (Palantir, Anduril, internal ML systems, etc.) are using significantly more advanced AI tools than ChatGPT so that’s not really the “news” of this story.
The framing is.
The heart of the DoW’s award calculus is, according to the reporting and public statements, is the Pentagon’s position on AI law.
OpenAI is down to collaborate on war if the military follows existing law — including bans on domestic mass surveillance and requiring humans to make decisions about the use of force. That’s enough for Altman and friends. If the government follows existing law, then AI can be deployed responsibly, in defense applications, under those rules.
Anthropic is not so down. They argue that AI dramatically increases the scale and power of data analysis. Their concern, which is apparently not shared by the Pentagon, is that even if the government only collects data that is technically legal to gather — such as public social media posts or other publicly available information — AI can combine and analyze that data in ways that effectively create large-scale surveillance. Anthropic’s position is that the law was written for human investigators and systems, not for systems that can instantly process billions of data points.
In other words, the law is insufficient.
Considering that the law is also insufficient to prevent mass surveillance in human systems, Anthropic’s point rings true. Now consider that the administration is also blocking state regulation of AI – so that insufficient federal law is really all we have.
When loopholes in the law are identified and brought to the attention of authorities and then left in place, unaddressed, it’s not unreasonable to speculate whether that means those authorities want the to loopholes to remain in place.
Is that strange for a tech domain that those authorities also describe as the main modern arms race?
Or is that kind of entrenched plausible deniability exactly what we should expect?
Please follow us on Gab, Minds, Telegram, Rumble, Gettr, Truth Social, Twitter
