House of Representatives bans the use of Copilot over security concerns

midian182

Posts: 9,748   +121
Staff member
What just happened? Microsoft has been dealt some more bad news related to its Copilot generative AI assistant: The US House of Representatives has issued a strict ban on congressional staffers using the tool over security fears. It now joins ChatGPT on the House's list of banned and restricted AI tools.

In guidance to congressional offices issued by The House's Chief Administrative Officer, Catherine Szpindor, and seen by Axios, it's stated that Copilot is "unauthorized for House use." The guidance adds that Copilot will be removed from and blocked on all Windows devices controlled by the House.

Szpindor said that the Office of Cybersecurity had deemed Copilot to be a risk to users due to the threat of leaking House data to non-House-approved cloud services.

Microsoft is planning to roll out a suite of government-focused tools this summer, which it says meet federal government security and compliance requirements. The Redmond company hopes these will address Congress' concerns.

"We recognize that government users have higher security requirements for data. That's why we announced a roadmap of Microsoft AI tools, like Copilot," a Microsoft spokesperson told Reuters.

Szpindor's office said that the guidance applies to the commercial version of Copilot, adding that the government version will be evaluated when it becomes available and a determination regarding its use will be made at that time.

For all the hype over generative AI, plenty of large companies and organizations have banned the use of the most popular tool of them all, ChatGPT. Amazon, Apple, Samsung, the World Economic Forum, Bank of America, and Verizon are just some of the big names that have implemented total bans or restrictions on OpenAI's generative AI among employees. House staffers were restricted from using the paid-for version of ChatGPT last year, limiting inputs to non-sensitive data, while use of the free version was banned altogether.

Microsoft has been aggressively pushing Copilot recently, launching a Pro subscription, quietly installing the Copilot app on Windows PCs, and releasing the new Surface Pro 10 and Surface Laptop 6 with dedicated Copilot buttons. However, according to a new report, Copilot for Microsoft 365 customers are complaining that it isn't as good as ChatGPT, despite being built on top of the same technology. Microsoft believes the issue lies with people who aren't using Copilot correctly or don't understand the differences between the two products.

Masthead: NATO North Atlantic Treaty Organization

Permalink to story.

 
If the group of people that have the most to hide wants to ban Copilot, then it must be really bad.
 
Not surprised. We all got a good laugh at our engineering guild meeting this week, as we watched copilot make up names for attendees and presenters by splicing first and last names together at random. 1/4 of the presenters and half the attendees apparently didn't appear at the meeting, and some folks who weren't presenters were named as if they were.

The tool is a joke. It actually had the invitee list as an attendee boundary and still managed to mangle roll call.
 
The clowns in congress did something right fro once. I'm hoping their are pirate builds of Windows 11 that remove copilot as well as TPM BS.
 
AI is not ready for prime time but companies are rushing into it so they can gloat.
They are in a rush because they have spent a lot of money on AI, and are trying to ride on the hype train to make money. While you get a free version, the one that they are trying to sell is the paid version.
 
Too bad. House legislation seemed like one of the very few areas where current AI might be able to outperform humans, or at least the ones currently doing that job.

Bigger picture, if this is the start of the House starting to wise up that we may need an overhaul of how devices are allowed to collect and share user data, good on them.

I feel like EFF or a similar organization should start a project where they shop from all available data brokers to try to find overly invasive data about every house member & staffer, and start posting it to make that point of what's out there. I wouldn't normally want to embarrass anyone, but as you can absolutely count on every one of our enemies (and allies) already having this same information, we might as well bring it to our representatives attention so they are aware of, and become motivated to fix, the privacy holes that are affecting not just them, but basically everyone else too.
 
Too bad. House legislation seemed like one of the very few areas where current AI might be able to outperform humans, or at least the ones currently doing that job.

Bigger picture, if this is the start of the House starting to wise up that we may need an overhaul of how devices are allowed to collect and share user data, good on them.

I feel like EFF or a similar organization should start a project where they shop from all available data brokers to try to find overly invasive data about every house member & staffer, and start posting it to make that point of what's out there. I wouldn't normally want to embarrass anyone, but as you can absolutely count on every one of our enemies (and allies) already having this same information, we might as well bring it to our representatives attention so they are aware of, and become motivated to fix, the privacy holes that are affecting not just them, but basically everyone else too.
What difference does a LLM make from a Legislator when the companies controlling them can tweak them to their own political biases should they want it to?
 
Its so important to help keep people free of Big Tech companies who want to control everything you do...
This is a good move, but the AI still needs to be removed completely, I have found traces of mine on my PC which was Windows 11 so I reinstalled 10
 
Back