When it comes to financial or legal matters (and this includes online banking) a small dedicated hardware element for signing fingerprints is all that's ever been required. Anything more is an overreach.
No, this is a misunderstanding of what a TPM is.
A TPM is a secure element inside your computer, similar to the chip running your credit and debit card. That's it. Without you using it (i.e. your OS or an application you installed asking it to do something), it's exactly as dangerous as a blank chip card in your house that you don't use and didn't open any account for.
If you don't want anybody to talk to it, don't install applications or OSes on your computer that do things you don't want. You have full control over that! Not running software that's not acting in your own best interests is generally good practice anyway, TPM or no TPM.
> [...] a small dedicated hardware element for signing fingerprints is all that's ever been required [...]
You might be happy to hear that that's exactly what a TPM is, then!
As you say, a TPM alone can't do much of anything and doesn't pose much of a threat. Of course expanding the acronym - Trusted Platform Module - is a bit of a giveaway. They were always fully intended to serve as the root of trust for much more nefarious things.
Conversely, DRM is alive and well on almost universally TPM-less devices.
By the way, all of your comments in this thread end up dead – I had to vouch for them to be able to answer. Not sure what’s up with that.
Because the immediate next step after locking devices down is profit extraction from users.
Do you think Apple would have been able to maintain their App Store margins absent device control?
There's simply too much incentive for abuse.
The only way I personally could see supporting them is if there were first a legislative requirement that trusted modules always be user-modifiable.
I think it's fair to assume that in a world in which almost every device supports attestation and makes it available to any service provider by default, without giving users an informed choice to say no or even informing them at all, service providers are much more likely to provide access exclusively to attestation-capable clients.
That, in turn, has obvious negative consequences for users with devices not supporting attestation (whether out of ideological choice, because it's a low cost device and the manufacturer can't afford the required audits and security guarantees etc.): Sure, these users will always be able to just refuse to transact with any service provider requiring attestation.
But think that through: We're not only talking about Netflix here. At what availability rates of attestation will decision makers at financial institutions decide that x% is good enough and exclude everybody else from online banking? What about e-signing contracts for doing business online? What about e-government services?
I am at the same time excited about the new possibilities attestation offers to users (in that they will be able to do things digitally that just weren't economically feasible for service providers, since they often have to cover the risks of doing so) as I am very wary of the negative externalities of a world in which attestation is just a bit too easy and ubiquitous.
In other words, the ideal amount of general purpose attestation availability is probably high, but significantly below 100% (or, put differently, the ideal amount of friction is non-zero). Heterogeneity of attestation providers can probably help a bit, but I'm wary of the inherent centralizing forces due to the technical and economical pragmatics of trusted computing.