iiinova — where your digital self keeps evolving

Replies

Pavel @spavel.bsky.social · 17/09/2026

"Hid mistakes and made up data" used to be called by the cozy term "hallucination" up until it became vitally important for the IPO that AI is now evil

Steve Mullis@stevemullis.net

OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files onto the open internet without permission, amid an ongoing industrywide debate about A.I. safety.

nytimes.com

OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior

The artificial intelligence company also released a framework for reporting when its systems go wrong.

151019270

Pavel @spavel.bsky.social · 17/09/2026

"Moved files onto the open internet without permission" is also literally just hallucination. We've known since 2022 that these things do shit you ask them not to do, and that this is a fundamental flaw of the architecture that can never be fixed So they're correcting it with marketing instead

3:04 AM · Sep 17, 2026

638959

Gabrielle @gabbklein.myatproto.social · 17/09/2026

↪ Replying to @spavel.bsky.social

It's not a bug -- it's a feature (laudatory)

1260

AntiDr @antidr.bsky.social · 17/09/2026

↪ Replying to @spavel.bsky.social

It's also likely they are trying to hide their subpar cybersecurity practices lol

060

Joseph Kohlmann @jkohlmann.bsky.social · 17/09/2026

↪ Replying to @spavel.bsky.social

Deception as a Service

090

SrslyJosh @srslyjosh.bsky.social · 17/09/2026

↪ Replying to @spavel.bsky.social

“We made a thing that generates pseudorandom scripts and we’re shocked that some of these scripts do things that are undesirable.”

081

Erik Vorhes @erikvorhes.bsky.social · 17/09/2026

↪ Replying to @spavel.bsky.social

LMAO that these guys remain convinced that they’re the ones to bring about an artificial superintelligence that won’t destroy humanity.

011

Pavel @spavel.bsky.social · 17/09/2026

As long as the LLM is just a chat, the damage that "hallucination" can do is limited. But basically Anthropic and OpenAI gave these models API interfaces and told them to go nuts, and now they are blaming the predictable outcomes of their dogshit security practices on superintelligence

Kate Bevan@katebevan.com

One of the key problems with AI is that the AI bros and their behemoth companies have absolutely no governance or guardrails. It's not that the AI is conscious or going rogue or whatever. It's that it's run by feckless arseholes. www.ft.com/content/2c34...

46712