Google, Pentagon Discuss Classified AI Deal

An anonymous reader quotes a report from Reuters: Alphabet's Google is negotiating an agreement with the Department of Defense that would allow the Pentagon to deploy its Gemini AI models in classified settings, the Information reported on Thursday, citing two people with direct knowledge of the discussions. The two parties are discussing an agreement that would allow the Pentagon to use Google's AI for all lawful uses, according to the report.

During the negotiations, Google has proposed additional language in its contract with the department to prevent its AI from being used for domestic mass surveillance or autonomous weapons without appropriate human control, the Information reported. The Pentagon will continue to deploy frontier AI capabilities through strong industry partnerships across all classification levels, a Pentagon official said, without confirming any talks with Google.

Re:"lawful uses"

[lucifuge31337]

It's like magats can't understand that two thing can be bad. And instead it's this crybaby "BUT WHATABOUT!?!?!"

"But they did it first" is not the defense you seem to think it is. People who behave badly should face consequences, and this is a thing MOST people agree on. You're just too far gone to know that - you're as bad as the lunatic left, but at least they don't have a literal cult leader.

Re:

[lucifuge31337]

But this time around the slavish anti-intellectual followers are outing themselves quite publically.

Re:

[anonymouscoward52236]

It won't be an autonomous weapon, but it will start wars, ask for straits to be opened, and then close straits at the threat of violence. All automated!

Uh oh!

[Local ID10T]

Google has proposed additional language in its contract with the department to prevent its AI from being used for domestic mass surveillance or autonomous weapons without appropriate human control

Google is about to be declared a threat to national security! Blocked from doing business with anyone who does business with the US Government! OMG! Supply chain risk! Terrorists! /s (Hey, it happened to Anthropic...)

Re:Uh oh!

[CrankyFool]

Not really. Google proposed these limitations, letting them claim proudly "hey, we proposed this!"

They didn't require these limitations, so my suspicion is Pentagon will say no, the deal will be classified so nobody will know, and Google will get the credit they want for "trying."

Re:

[functor0]

Even if these limitations were accepted into the contract, there's still the "with appropriate human oversight" caveat that's practically worthless.

Re:Uh oh!

[Local ID10T]

For those that missed it... Anthropic proposed to code the restrictions into the model itself. Other companies have proposed putting these restrictions in their contract -but lack any way of verifying or enforcing them beyond taking the government's word for it.

So tech companies are now the source of ethics

[Botched]

This is not going to last.

Department of War almost checked in middle east

[oumuamua]

Expect for one guy

With the decisive support of one Democrat—Rep. Jared Golden of Maine—the Republican-controlled House of Representatives on Thursday voted down a war powers resolution aimed at ending President Donald Trump’s illegal assault on Iran

https://www.commondreams.org/n... [commondreams.org]
And Bernie Sander's resolution also held up:

The Dam Breaks: Democratic Senators Overwhelmingly Reject Arms Sales to Israel Despite their defeat by Senate Republicans, bills seeking to block arms sales to Israe

If it's classified, you can hide it from employees

[david.emery]

Only those with Need to Know/Read into the program will know the details, beyond what's being reported. So lots of Goole employees who might object to this will remain ignorant of what management is actually up to.

I suspect a lot here see that as a significant problem. (But in the grand scheme of things, I'm not losing sleep over this one.)

Prediction: Virtual President

[epicbread]

Similar to Zucky's attempt to make a virtual himself, there could be could be made a virtual president, trained on any given president's thought process and style, in order for a secondary leader assigned such a helm, be coached in that degree of thought process, in the absence of the actual.

Let me guess where the government

[Growlley]

defines lawfull in secret courts & it's unlawfull to even disclose the issue is being adjudicated?

Who ends up controlling this AI?

[hcs_$reboot]

The real issue is who ends up controlling this AI
It’s not that AI is inherently more intelligent than humans, but that the people using it will trust it completely.

Haaa haaaa ha!

[Mirnotoriety]

> Alphabet's Google is negotiating an agreement with the Department of Defense that would allow the Pentagon to deploy its Gemini AI models in classified settings ..