4 ms·

> Try convincing everyone that grammarly is a keystroke logger. Funny how it's not the opposite viewpoint that is scrutinized. Prove how it's not a keystroke l

> Try convincing everyone that grammarly is a keystroke logger. Funny how it's not the opposite viewpoint that is scrutinized. Prove how it's not a keystroke logger, then Grammarly can be deployed. Until then, it's assumed to be malicious. That's how people treat for example ChatGPT as well.

roenxi 3y ago
I'm not sure if there is official research on this, but the 'opposite viewpoint' sounds expensive and more prone to failing in practice. Software companies generally struggle to get results even with a willingness to use any tool or practice that might improve productivity. It will only get worse if they add paperwork. And it isn't very risk-based. In most cases, the keylogger can be assumed to be entering via the OS, probably based on some NSA/GRU/Chinese MSS exploit being used for industrial espionage. Grammarly is a bit more niche. It is a threat. One day it will be an acute one, perhaps. But the risks and opportunities favour recklessness.
PeterStuer 3y ago
Yet those same people happily use Office 365 without a second thought.
- TeMPOraL 3y ago
  Microsoft is a serious corporation, i.e. a fat target for lawsuits. That makes them more trusted by default than random fly-by-night startups. On top of that, any medium or large business using Microsoft tools will have a contract signed with them, one that covers SLA and data handling policies. In fact, this is pretty much the entire value proposition of Azure deployments of OpenAI models - from a corporate POV, OpenAI itself is a fly-by-night startup that can disappear any day, while leaking the company secrets your employees gave it all across the Internet. In contrast, Microsoft's ChatGPT (the very same models) comes with an SLA and data handling policies suitable for dealing with corporate secrets (with whole stack of audits and certifications behind them). That's why corporate will tell you to not use ChatGPT, but if you're lucky, they'll give you access to a copy of it on Azure.

roenxi 3y ago

I'm not sure if there is official research on this, but the 'opposite viewpoint' sounds expensive and more prone to failing in practice. Software companies generally struggle to get results even with a willingness to use any tool or practice that might improve productivity. It will only get worse if they add paperwork. And it isn't very risk-based. In most cases, the keylogger can be assumed to be entering via the OS, probably based on some NSA/GRU/Chinese MSS exploit being used for industrial espionage. Grammarly is a bit more niche. It is a threat. One day it will be an acute one, perhaps. But the risks and opportunities favour recklessness.

PeterStuer 3y ago

Yet those same people happily use Office 365 without a second thought.

TeMPOraL 3y ago
Microsoft is a serious corporation, i.e. a fat target for lawsuits. That makes them more trusted by default than random fly-by-night startups. On top of that, any medium or large business using Microsoft tools will have a contract signed with them, one that covers SLA and data handling policies. In fact, this is pretty much the entire value proposition of Azure deployments of OpenAI models - from a corporate POV, OpenAI itself is a fly-by-night startup that can disappear any day, while leaking the company secrets your employees gave it all across the Internet. In contrast, Microsoft's ChatGPT (the very same models) comes with an SLA and data handling policies suitable for dealing with corporate secrets (with whole stack of audits and certifications behind them). That's why corporate will tell you to not use ChatGPT, but if you're lucky, they'll give you access to a copy of it on Azure.