I had mixed feelings upon reading this news article, wherein Signal Foundation CEO draws attention to the extremely leaky problem of LLM agents's pervasive access to user data.
On the one hand, it's good to know that Signal leaders are aware of the leaky nature of the devices they force users to use to start using Signal.
It sucks, though, that they perceive this as an LLM issue.
If the LLM, when granted access to user data, amounts to a privacy, safety and even security problem, the operating system that has access to the data and manages access to it poses the same problem in as much as it is also leaky and not under user control.
IOW, Signal leaders understand the problem and know that the privacy, safety and security properties they purport to offer are invalidated by the leaky, rotten infrastructure they build upon and force all of its users to go through at least once.
Why do they go through all this trouble then?
So blong,