Silicon Valley culture is deeply, deeply adverse to truly competitive markets because they are inevitably low-margin, meanwhile China's companies will work into the ground to shave off a fraction of a %. The age of software as a high-margin business is coming to a close and the fat, lazy and bloated US software mega-corps are going to really struggle.
The 'global free internet' was always a second order effect of American hegemony, as that continues to break down so will the internet as we know it. You're seeing the same with banking as SWIFT gets more regional competitors.
There are parallels here to the way much of Africa fixed land-lines phones and went straight to mobile. Which has also been powered by Chinese hardware.
Arguably if you consciously choose to invest on a stock there should be no protection or obstacle other than regulated controls. For indices I see the point of extra protection. Where it gets murky is when you look at retirement funds, where you still have no control, but that should be regulation on the funds themselves rather than the stock.
Given Titan was an entirely preventable disaster borne of arrogance and stupidity it seems an apt analogy for the frontier labs 'sandboxing' efforts so far.
If they genuinely believe these models are going to obliterate the human race their utterly cavalier attitudes is be grounds for immediate government shutdown/takeover and arrest of everyone involved. If this was nuclear or biological research work none of this would be tolerated.
Their recent update on this situation was hilarious.
> Our [new] approach to containment starts with a paramount requirement: the limits on what an agent can access must hold even when the agent tries to circumvent them. We define a strict security boundary around the resources the agent is authorized to use and design the surrounding infrastructure to enforce that boundary—even when agents make sustained, sophisticated attempts to breach it.
Oh! So we’re digging a moat again, but this time an extra-super-double-good moat. With alligators. Well that’s that sorted then!
Maybe we can use AI agents to patrol the boundary for the AI agents. After all, the only thing that stops a bad guy with an AI swarm is a good guy with an AI swarm…
Ah! Of course:
> Teams of frontier-model agents were tasked with finding vulnerabilities and misconfigurations that could allow them to take unauthorized actions […] Accordingly, we’ve made high-compute red-teaming part of the safety cases we use to evaluate whether our frontier training and evaluation runs can proceed securely.
We added a strip of reflective tape which confuses its navigation sensor, we never expected it to run its sweeper motor at 10% power and push a pile of debris over the strip. Our bad.
Data visualisation can never be neutral for the same reason photography can never be neutral - you cannot do it without making endless decisions about framing even if you try and capture your subject ‘neutrally’. What you include, what you don’t, whether your axies focus on the charge range or full numeric ranges and so on. See Susan Sontag’s On Phototography for a deeply considered read on this.
The total failure of PurpleBricks attempt to disrupt estate agents in the UK is a good example of this. Turns out people had no idea of the shit-shovelling, chivvying and smoothing-over agents do to help sales actually close.
Sounds like some PM is jealous of Whoop and trying to launder the cost and boost adoption of their unwanted AI feature/subscription behind a heavily used existing feature.
Look at spaces where ripoffs/copying are already rampant like mobile games. The result is a slop avalanche, rending the whole space commercially unviable for anything other than the most bottom-of-the-barrel dark-pattern shitware laden with scam ads and spyware because users don't care and won't pay.
That may well be the future for most commercial software (and software engineers) as margins collapse and AI tokenomics moves more power from labor to capital.
reply