US Space Force Pauses Generative AI Use Based on Security Concerns

US Space Force has temporarily banned the use of web-based generative AI tools and so-called large language models that power them, citing data security and other concerns, Bloomberg News reported Wednesday, citing a memo. From the report: The Sept. 29 memorandum, addressed to the Guardian Workforce, the term for Space Force members, pauses the use of any government data on web-based generative AI tools, which can create text, images or other media from simple prompts. The memo says they "are not authorized" for use on government systems unless specifically approved.

Chatbots and tools like OpenAI's ChatGPT have exploded in popularity. They make use of language models that are trained on vast amounts of data to predict and generate new text. Such LLMs have given birth to an entire generation of AI tools that can, for example, search through troves of documents, pull out key details and present them as coherent reports in a variety of linguistic styles. Generative AI "will undoubtedly revolutionize our workforce and enhance Guardian's ability to operate at speed," Lisa Costa, Space Force's chief technology and innovation officer, said in the memo. But Costa also cited concerns over cybersecurity, data handling and procurement requirements, saying that the adoption of AI and LLMs needs to be "responsible."

wut?

[CEC-P]

Well now how are they going to generate some "very convincing" pics of them ayyyy lmaos to send to the Mexican president? But seriously, what was a branch of the military going to do with generative AI? Write documentation? I know from experience that you can't use insecure methods for that or to do markups and keyword tagging. Model rendering? Same problem. I think they're banning something that barely existed in the first place in that particular workforce.

Re:

[CEC-P]

Type "space is racist" into Google and it'll answer your question. The results are pretty entertaining.

Re:

[garyisabusyguy]

I would expect AI to be useful is analyzing multiple input data streams like high resolution imaging and wireless comms for threat identification

Of course, you would not want it to have control of any weapons systems for a response, just notification to the chain of command that leads to some human's finger on a trigger or button

Re:

[ShanghaiBill]

I would expect AI to be useful is analyzing multiple input data streams like high resolution imaging and wireless comms for threat identification

Sure, but that's classification AI, not generative AI.

I have no idea what Space Force uses generative AI for, but a blanket ban seems stupid. There are likely some good uses. Leave the decision up to individuals and local commanders. Micromanagement is bad, especially in the military, where initiative and rapid decisions are critical.

Let politicians test-pilot

[Tablizer]

...rockets designed by AI.

Re:

[garyisabusyguy]

Oh for fucks sake Boris, just because you do not understand English well enough to get a joke, please do not attempt to interpret it for native speakers

Re:

[garyisabusyguy]

Boris is whistling past graveyard as Russia invades neighbors in friendliest ways possible... nothing to look at here

Re:

[Tablizer]

MitchBot is stuck in a reboot loop.

This is just common sense

[Thud457]

No government function should be posting sensitive information to tools on the web.

Except for the War Thunder forums [wikipedia.org], naturally.

Re:

[sysrammer]

Heh.

June 2022 ZTZ-99 tank Classified Tungsten penetrator A user posted an image of a Chinese DTC10-125 tungsten penetrator on the War Thunder forums, sitting atop a document outlining the projectile's specifications. Similar to previous instances, the post was quickly removed by forum moderators, with them citing the fact that the materials in question are classified in China.[62]

Assuming that the leaked source was Chinese, I wonder what this guy's social score is nowadays.

Misdirection

[spaceman375]

They have enough money to buy their own, but you can bet money that they have multiple projects going already, from various sources as well as in house.

this is only funny because Space Force

[toxonix]

So did every company and government agency with any data they don't want to share with a third party. Seriously, Space Force has a 30 Billion dollar budget. "When describing space threats, it is important to account for two kinds: first, threats from space assets and second, threats to space assets" Wow, mind-blowing. So one space asset can pew pew with lasers, and another asset can GET pew pewed with lasers. I hadn't considered that aspect of space wars.

Re:

[cusco]

After only four years Space Farce has a budget larger than NASA. I guess that's to be expected, since the Pentagram spent more last year than all the NASA budgets since its foundation COMBINED, including the entire Apollo program.

we also took Global Thermonuclear War off the game

[Joe_Dragon]

we also took Global Thermonuclear War off the game list