Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
×
AI Technology

National Archives Bans Employee Use of ChatGPT (404media.co) 10

The National Archives and Records Administration (NARA) told employees Wednesday that it is blocking access to ChatGPT on agency-issued laptops to "protect our data from security threats associated with use of ChatGPT," 404 Media reported Wednesday. From the report: "NARA will block access to commercial ChatGPT on NARANet [an internal network] and on NARA issued laptops, tablets, desktop computers, and mobile phones beginning May 6, 2024," an email sent to all employees, and seen by 404 Media, reads. "NARA is taking this action to protect our data from security threats associated with use of ChatGPT."

The move is particularly notable considering that this directive is coming from, well, the National Archives, whose job is to keep an accurate historical record. The email explaining the ban says the agency is particularly concerned with internal government data being incorporated into ChatGPT and leaking through its services. "ChatGPT, in particular, actively incorporates information that is input by its users in other responses, with no limitations. Like other federal agencies, NARA has determined that ChatGPT's unrestricted approach to reusing input data poses an unacceptable risk to NARA data security," the email reads. The email goes on to explain that "If sensitive, non-public NARA data is entered into ChatGPT, our data will become part of the living data set without the ability to have it removed or purged."

National Archives Bans Employee Use of ChatGPT

Comments Filter:
  • Just "ChatGPT"? (Score:4, Interesting)

    by cstacy ( 534252 ) on Wednesday May 01, 2024 @05:33PM (#64440180)

    Are they blocking Bing, too? What about Google and other search engines that function by using a LLM-AI

    And that's just about leaks. Is the official keeper of Government business and historical records considering these "AI" threats in the other direction: importation of information from untrustworthy digital agents, contaminating and corrupting the official records?

    My first experience with ChatGPT when it was first released was that it casually mentioned, umprompted and seemingly out of context, that Hillary Clinton was the President, having beaten Donald Trump. And I thought to myself, "I know that's not true, but if I was a 4th grader in History class and this got slipped in there, I'd just believe it!"

    • by narcc ( 412956 )

      if I was a 4th grader in History class and this got slipped in there, I'd just believe it!

      This is exactly why these things should not be used as private tutors. Given the way the function, they are guaranteed to give incorrect information, as more and more people are discovering in embarrassing and expensive ways.

      contaminating and corrupting the official records

      I see it like the wikipedia "citogenesis" problem. AI generated nonsense gets introduced into what should be a trustworthy source gets copied other places, including other ostensibly trustworthy sources, lending it further credibility.

      AI generated content is pollution.

    • Re: (Score:1, Offtopic)

      by DarkVader ( 121278 )

      If we'd eliminated the anti-democratic electoral college like we should have over a century ago, she would just be finishing up her 2nd term.

      She won by 3 million votes. But unfortunately land has more of a vote than humans in this country.

      • Re: (Score:1, Insightful)

        by Anonymous Coward

        The United States of America is, by design, a federal republic. It is supposed to govern sovereign states, not their people.
        The 16th, 17th, and 23rd Amendments have all eroded this distinction, but for now states still choose electors.
        Getting rid of the electoral college, such that president is popularly elected by all US voters, would mean that only the interests of the big cities matter.

    • Searching the Internet is okay - it's when you paste a load of your company secrets into the prompt box and say "summarise this for me" or "write something like this" or whatever else that the problems start.

      If it's not a self-hosted, fully offline LLM, you should assume the owners of it are using the prompts you type in to train the model. The more you paste into the box, or the more questions you ask, the more likely your transcript will get their attention. After all, they want to make their product do w

  • Assume you are being scanned by ai on every website or app.
  • Considering how many people are secretly using workarounds to access WhatsApp despite it being banned in their country [slashdot.org], I have no faith that the National Archives blocking access to ChatGPT will make much of a dent in the problem.
    • Re: (Score:3, Insightful)

      by Anonymous Coward

      This ban is for government provided laptops that have access to NARA data. Try to install ChatGPT on a government laptop and get a quick visit from your supervisor. Try to install an unauthorized VPN on your laptop and get the same.

      There is actually a large list of software that is not authorized on government computers such as games, TikTok, WeChat, Kaspersky anti-virus, remote logon, bitcoin miners, and others. If you want to use those products, do it on your personal computer not your government comp

The trouble with computers is that they do what you tell them, not what you want. -- D. Cohen

Working...