vayup

OpenAI models prove ETH is false. BTC it is then!

</joke>

Agree with everything on your post. I have to say as a frontier lab employee, it hurts a little to be compared with McKinsey, even as an analogy. :-)

> However, the level of social media amplification around this incident has brought scrutiny that is out of proportion to SSL/TLS’s place in our business

Translation: We are exiting this business because people are being mean, and we don't get paid enough.

WTF!

They also say: > We believe an organization's character is shown in how it responds when something goes wrong

I totally agree with this. I will judge their character based on this: No apology for wreaking havoc on the world, no admission of fault, whining about people getting pissed at the monumental incompetence, saying that they are exiting the business because "sour grapes" rather than admit they have lost the trust within the community. <sigh>

For those who think this is about training data: If that is the concern, they would sanitize the training data, like they do for thousand other things. No chance in hell that they would rely on users meticulously following their usage policies for the quality of their traning data.

Frontier AI folks may be crazy, but not crazy enough to believe people read usage policies :-)

I bet much of the savings is from reducing theft. 50 percent is too much for transmission losses.

In the industry, transmission and distribution losses are called T&D losses. Pretty much all unaccounted losses will be written off as T&D losses. Often the right expansion for T&D losses is Theft and Decoity losses :-)

They also effectively said that military contractors and their subcontractors cannot use Anthropic - that's a significant part of the economy. I mean technically, they can use Anthropic for non-military work. But in practice, they will choose not use it at all, since it's a better approach to show compliance.

Makes sense. Thanks.

Hm... not normal in my experience. Not enabling a config is not a vulnerability in itself. If not enabling something means a security guarantee is broken (Eg: videos are accessible) then it is a vulnerability, and typically included in VDP, atleast VDPs that are in good faith.

It's reasonable to not have a VDP for the reasons you mentioned. But not reasonable to have a useless one just so it appears they have a VDP.

I am a bit confused at people looking at this negatively. If Nvidia believes AI demand is extraordinarily high, and that it requires a large capital outlay to serve that capital, isn't this what they should be doing?

MITM is a feature now.

And yet, Amazon Prime is inspired by Costco membership.

> They've got, ballpark, $5t to $10t to make back in the next 5 years, or the hardware buildouts will start getting written down.

Depreciation and write-offs are about accounting models. Hardware will still be running after five years and still be making money. They may not be as efficient as the new hardware, but they will still be making real money even though they are valued at $0 in the books.