Microsoft Temporarily Blocked Internal Access to ChatGPT, Citing Data Concerns
wsj.com
wsj.com
> The decision to block ChatGPT, which is made by Microsoft-backed startup OpenAI, was made and communicated by the information-technology department, according to a person familiar with the matter. Management was caught by surprise by the move and later reversed the decision, the person said.
"Security concerns"... decision made by Microsoft internal IT, and then overruled by management.
Unsurprising, really; ChatGPT is not what I would consider a secure system in any way.
> "We were testing endpoint control systems for LLMs and inadvertently turned them on for all employees," a spokesperson said. They added: "As we have said previously, we encourage employees and customers to use services like Bing Chat Enterprise and ChatGPT Enterprise that come with greater levels of privacy and security protections."
Keep in mind Microsoft themselves develop and sell endpoint control systems, including lists of "bad" sites to optionally block. Microsoft also dogfoods their own stuff. It isn't difficult to imagine how this occurred.
I wonder what criteria they employ to determine how "bad" a site is.
Here is what they already allow you to block:
- Adult Content/Gambling/"Tasteless"/Nudity/Violent
- High bandwidth: Video streaming, image sharing, P2P, etc
- "Legal liability": Inc. "hacking sites", hate groups, pirate software, illegal drug, school cheating, self-harm, et al
More information here[0]. You can enable at the parent or each child category. They'll likely add a new parent category for "AI" et al.
[0] https://learn.microsoft.com/en-us/microsoft-365/security/def...
A lot of financial companies have an allowlist approach to their firewalls, so if it's not on the list it needs to be blocked.