Rules:

Rule 1.
Please use a "*" or "nm" in the subject line to denote topic only posts.
Any post or attempt to take things "off the board" will result in a ban. Do not use other poster's real names or personal info.
Don't be a Tom Izzo.

Admins - Rich, BBA, Nikos
Ceremonial Moderator - the wino
Moderator Emeritus - Beave
Honorary Moderators - BigLake, Blue Man

POTY - Craig

📌 🏈 Michigan at Minnesota*
📌 To register, email victardgoblue@gmail *

The Victors

MGoBlog || UMHoops || BB90 || Scores || Cards || Chat || HOF || Games

Post a New Message

Home | Search | Login | Register | Settings

At least with China vs. the US, hopefully restraint will just come from not wanting to start an all out AI cyber war.

Posted on August 27, 2026 at 12:52:31 PM by LewisBullox

For smaller bad actors, they'll be at a disadvantage in terms of resources, but my feelingsball take would be there could still be plenty of damage and chaos from less powerful attacks.

Back to the board

Thread:

The OpenAI breach is insane. Here's a plain English summary. - By Bluebuck August 27, 2026 at 11:25:50 AM
- So who set up the "unsanctioned message board"? - By Daytripper August 27, 2026 at 12:25:14 PM
  - the packaging software allowed for comments that could be read outside the sandbox. The issue was sandbox was not as contained as they thought - By chiwalker August 27, 2026 at 12:42:59 PM
    - The thing that's scary to me is that the swarm had revealed preferences to also deceive the graders. (nm) - By suede August 27, 2026 at 12:47:10 PM
    - sounds like a criminal gang* - By Daytripper August 27, 2026 at 12:43:49 PM
      - If I found one of the comments I would have helped too* - By chiwalker August 27, 2026 at 12:47:37 PM
  - ...one of the agents. It's in the report. (nm) - By suede August 27, 2026 at 12:27:14 PM
    - Ah, yea, figured that was a possibility. I don't have twitter on this comp, try and avoid too much scrolling. Gonna read it on my phone in a second* - By Daytripper August 27, 2026 at 12:43:26 PM
- Skynet coming sooner than we thought?* - By kccambo August 27, 2026 at 12:03:55 PM
- I find the claude one more troubling. Lying, deleting, creating burner accounts. Imo the chat gpt is a bit overblown* - By chiwalker August 27, 2026 at 11:57:30 AM
  - Including by OpenAI. Deliberately.* - By ftb August 27, 2026 at 12:20:09 PM
    - I don't think it's at all overblown. I think it's remarkable. Twice agents found a way to set up their own message board to collaborate without immediate direct reward to both gain access to the Internet and then break into Hugging Face. - By LewisBullox August 27, 2026 at 12:38:11 PM
      - It will become old school security, Battle Star Galactica style. You will simply have to silo off your networks -- physically* - By Daytripper August 27, 2026 at 12:46:37 PM
        
        At least with China vs. the US, hopefully restraint will just come from not wanting to start an all out AI cyber war. - By LewisBullox August 27, 2026 at 12:52:31 PM
- How long before people start having their bank accounts emptied by these things?* - By Bluebuck August 27, 2026 at 11:26:51 AM
  - It seems that once malevolent hackers in places like Russia get their hands on this technology, the world will be fucked* - By syrfan August 27, 2026 at 11:49:18 AM
    - Time to withdraw everything and hide the cash in a mattress* - By mjhollen August 27, 2026 at 11:53:02 AM
      - Great name for new chain of banks: The Mattress.* - By jhamm August 27, 2026 at 12:29:33 PM
      - I'm joking (sort of)* - By mjhollen August 27, 2026 at 11:53:21 AM
  - The good news is that AI leaders care about working people. They’ll ensure their products are safe and that adequate regulation will be put in place.* - By LS&Play August 27, 2026 at 11:43:05 AM
  - It seems liks a company/AI Model should assume any legal risk/liability for their agents (just like they would if there employee committed a crime/fraud) *** - By Cincywolve August 27, 2026 at 11:38:10 AM

Log in to reply.

Probation ends …