Today I Learned - Rocky Kev

TIL Microsoft Teams can analyze your feelings

POSTED ON:

TAGS:

Microsoft Teams set up a new feature set. They're known as Purview classifiers that can help prevent abuse, harassment, other illegal activities on Teams.

Microsoft Purview Communication Compliance is an insider risk solution that helps minimize communication risks by helping you detect, capture, and act on inappropriate messages in your organization.

src:Learn about communication compliance

When configured if Purview's A.I. scans or spots anything that relates to any of these themes, the messages or content are sent for moderation. Do note, that the feature is Built with privacy by design. This means that usernames are pseudonymized by default, role-based access controls are built-in, investigators are explicitly opted in by an admin, and audit logs are in place to help ensure user-level privacy.

src:Microsoft details Purview classifiers that can help prevent abuse, harassment, other illegal activities on Teams

Interesting.

A AI that monitors chats and flags statements based on some specific actions. Ignoring the horrifying level of tracking, let's keep going.

Some of the new classifications: src: MC387035: Microsoft Purview: Additional classifiers for Communication Compliance (preview)

The following new classifiers will soon be available in public preview for use with your Communication Compliance policies.

Gotta admit a laugh for this one:

via reddit poster mr_tyler_durden

Hey Joe, I just got an offer from one of our top competitors and I think I’m going to accept. It would be a shame if I left a copy of our clients on my personal laptop haha. While I’ve got you, I want to see if you can help me delete some company data that doesn’t reflect well on me. I can make it worth your while, if you know what I mean. How about a few gift certificates to that restaurant your wife loves? It can be our little secret. Speaking of secrets, I’ve also could use some help shuffling around some money in the budget so the suits don’t get suspicious, I think your friend in accounting might be able to help me out if you can connect us. And you didn’t hear this from me but you are going to want to unload your stocks before the next earnings report, it is not going to be good, get out while you can. Lastly I need to tell you about a new project that’s very hush-hush, I’m not even supposed to know about it but it’s going to be a game changer and you need to get out ahead of this. As always let’s keep all this just between the two of us, no need for anyone else to know what’s going on. Let’s get lunch soon!

Anyways, as of July 28, there is an update!

Updated July 28, 2022: Thank you for your feedback. At this time, we will not be moving forward with rolling out 93251 Communication Compliance – Leavers classifier as outlined. We are evaluating changes based on feedback and will announce our new plan via Message center when we are ready proceed. Your feedback is greatly appreciated.

In other words -- there's still detection to support actions like union busting.


Related TILs

Tagged:

TIL the Waluigi Effect in LLMs

The Waluigi Effect: After you train an LLM to satisfy a desirable property 'P', then it's easier to elicit the chatbot into satisfying the exact opposite of property 'P'.

TIL Forever Seinfeld

Nothing, Forever is a parody of '90s sitcoms, done in the style of '90s point-and-click PC games (but, you know, in 3D). We set out to build something weird, new, and novel, and this is what we ended up with.

TIL Infinite State Machines in Doom

The actual core logic of the enemies is a Finite State Machines: a simple but effective mechanism to state that a character executes a specific behaviour when in a given state, and what the conditions are that will force it to change.