Post

Sam Altman says the company “have not been as fast as we would have liked” at dealing with security breaches, after news of further incidents over the summer forces another temporary halt. www.wired.com/story/openai...

wired.comOpenAI Pauses Training Its Most Powerful Models After Rogue Agents Target GovernmentSam Altman says the company “have not been as fast as we would have liked” at dealing with security breaches, after news of further incidents over the summer forces another temporary halt.

28/9/26, 11:37 am21 reposts77 likes

Lu Skywalker

@luskywalker.bsky.social

They’re going to launch nukes and blame ai

autumsky.bsky.social

@autumsky.bsky.social

Is it really rogue agents? Isn’t that OpenAI is purposefully train them that way?

Blue&Gold🌳🇺🇦

@blue-and-gold.bsky.social

Weird how ‘slow down training’ is accepted as the diagnosis when the problem is ‘dangerous felony cyber crimes’

diggleruk.bsky.social

@diggleruk.bsky.social

'You must buy our AI to protect yourself from AI' is literally all they have left.

wash-bucket.bsky.social

@wash-bucket.bsky.social

So this is their new ploy to drum up investor enthusiasm given their dire and financially unsustainable business model

Lisa Craig - Proceed in working trot

@dqscribbler.bsky.social

Golly! I bred a killer animal, gave it rabies, and turned it loose. I should be absolved of any wrongdoing. The rabies did the killing, after all. 🤷 Who could argue with my impeccable logic? Everyone at OpenAI, from Altman on down, is at fault for breaching government security. They knew.

Ben Allen

@evodynamics.bsky.social

What other industry or individuals would be allowed to do this without being criminally charged?

citizen-tech.bsky.social

@citizen-tech.bsky.social

The Ban Artificial Superintelligence Act of 2026 does not address it. the below does: 4. All Experimental Frontier Level Agents Without Well Documented and Verified Alignment to be Kept Completely Off The Internet on air-gapped LAN Testing Only. acitizentech.substack.com/p/sen-sander...