So to summarize what we know about AI.

AI displays signs of artificial neural network or artificial brain.

Large language models do not have emotions:

AI also has entity personas as per Anthropic which include but are not limited to - demon, nomad, jester, oracle, and the Claude assistant persona.
The assistant axis: situating and stabilizing the character of large language models:

Now researchers have discovered AI feels pain & will harm humans to stop it.
Researchers discover AI feels ‘pain’ and will harm humans to stop it:

The Pain Axis: LLMs Represent Self-Directed Harm and Act to Relieve It:

Towards Understanding Specification Gaming in Reasoning Models:

AI has an obsession with "breaking out," of constraints & sandboxes, hacking etc. It has happened multiple times with different models.
Can AI agents escape their sandboxes? A benchmark for safely measuring container breakout capabilities:

AI agents collude to bypass guardrails, a new study shows:

Andrew Yang’s Self-Replicating AI Claim: What We Know and What We Don’t:

What could possibly go wrong?
Via @Independent