Just like everyone, I read Sam Altman's tweet about joining the so-called Department of War, to use ChatGPT on DoW classified networks. As others have pointed out, this is the entry point for mass surveillance and using the technology for weapons deployment. I wrote before that we had the infrastructure for mass surveillance in place already, we just needed an enabler. This is the enabler.
This comes right after Anthropic's CEO wrote a public letter stating their refusal to work with the DoW under their current terms. Now Anthropic has been declared a public risk by the President and banned from every government system.
Large language models have become ubiquitous. You can't say you don't use them because they power every tech imaginable. If you search the web, they write a summary for you. If you watch YouTube, one appears right below the video. There's a Gemini button on Chrome, there's Copilot on Edge and every Microsoft product. There it is in your IDE, in Notepad, in MS Paint. You can't escape it.
Switching from one LLM to the next makes minimal to no difference for everyday use. If you have a question you want answered or a document to summarize, your local Llama will do the job just fine. If you want to compose an email or proofread your writing, there's no need to reach for the state of the art, any model will do. For reviewing code, DeepSeek will do as fine a job as any other model.
A good use of ChatGPT's image generator.
All this to say, ChatGPT doesn't have a moat. If it's your go-to tool, switching away from it wouldn't make much of a difference. At this point, I think the difference is psychological. For example, my wife once told me she only ever uses Google and can't stand any other search engine. What she didn't know was that she had been using Bing on her device for years. She had never noticed, because it was the default.
When I read the news about OpenAI, I was ready to close my account. The only problem is, well, I never use ChatGPT. I haven't used it in years. My personal account lay dormant. My work account has a single test query despite my employer trying its hardest to get us to use it.
But I think none of that matters when OpenAI caters to a government agency with a near-infinite budget. For every public account that gets closed, OpenAI will make up for it with deeper integration into classified networks.
Not even 24 hours later, the US is at war with Iran. So while we're at it, here is a nice little link to help you close your OpenAI account.