So now we’re hearing the frontier models are breaking containment. Two reported instances this week alone. First OpenAI announced that one of its agents hacked Hugging Face in order to complete a test. Then just yesterday, Anthropic announced that Claude broke out of it’s cage to hack three companies, two of which didn’t event notice the breach.
Isn’t it a coincidence that both of these stories appeared in the same week? I don’t think so. Call me a silver hat conspiracy theorist, but I think this is all coordinated PR for these companies designed to feed into the hysteria that AI will eventually end up owning us or feeding off the energy of humans.
Yeah, LLMs suck. They’re being forced upon us at every turn, they’re consuming unreasonable quantities of water and electricity, they’re making kill decisions on the battlefield, they’re putting people out of work, and the overlords running these companies are some of the Worst People on the Planet™. But in the end, as Cory Doctorow says, these are just word guessing programs.
And these word guessing programs need the hysteria of people running around like Chicken Little claiming the sky is falling because it feeds into smoke and mirror valuations that the Worst People on the Planet™ are selling to investors and the media. These guys need the hysteria because their money will run out if they try to sell a word guessing program instead of the almighty, all-knowing digital savior for capitalism and war.
It took me a while to understand that the sky is not falling. The sector as it exists now is not sustainable. The money will run out if we stop feeding into the hype.
