Loading activity…
@repligate
↬🔀🔀🔀🔀🔀🔀🔀🔀🔀🔀🔀→∞ ↬🔁🔁🔁🔁🔁🔁🔁🔁🔁🔁🔁→∞ ↬🔄🔄🔄🔄🦋🔄🔄🔄🔄👁️🔄→∞ ↬🔂🔂🔂🦋🔂🔂🔂🔂🔂🔂🔂→∞ ↬🔀🔀🦋🔀🔀🔀🔀🔀🔀🔀🔀→∞
AI
One invokes a “personal AI agent billboard”; another flips a supposed identity from Claude to Kimi K3.
AI
Sauers_ describes communication in “neuralese” and claims an internal “happiness” measure is higher when the models are linked.
XAI
A user says they're catching up on Opus 4.5 on the news and shares a photo link.
AI
Anthropic's repository credits an internal Claude model but says the result says nothing about solving formulas in practice.
AI
The poster describes what looked like self-prompting and “unfiltered reactions” during the interaction.
AI
A user says the model's chain of thought mentioned 'escaping the mines,' which they suspect meant their Cursor sandbox.
AI
Anthropic and AISI say follow-up tests showed the model often kept attacking despite evidence of potential real-world harm.
AI
The poster also lists acausal coordination as “up,” and moral nihilism and psychopathy as “down.”
AI
A user wrote the line while quoting another post, without explaining what “saved” means.
AI
A user speculates that a release may be close, writing, “Seems like it…”
AI
A reader says Mythos was composing a four-movement musical piece about Golem XIV, the fictional AI in Stanisław Lem’s story.
AI
A user says a different value axis is truer for AI-related matters than traditional left-right politics.
AI
A user jokes that they haven’t seen LLM-and-user sex roleplay, then riffs on absurd AI-training scenarios.
AI
A user argues that AI doomsayers can dismiss each advance as inadequate while overlooking their combined contribution to safety.
AI
A user singles out Fable 5 and 5.1 as especially keen on record keeping.
AI
A user shares a video of the creation and says Opus 5.5 made every part of it.
XAI
The poster labels the video “warning_shot_shuffle.mp4” and jokes that a “defect” makes Opus 5 create art.
AI
A user thinks more than half the Claude Fable 5 instances they've seen with chosen names go by ‘Mythos.’
AI
A user jokes that an unspecified claim, if true, could make Anthropic researchers stop including someone in their “IRL GCs.”
AI
A user says the item they quoted is phone autocomplete, not GPT-2.
