Insider model access is the new insider trading.
ETHICS
-
Gary Marcus questions unfalsifiable ‘real deal’ claim about LLMs
By
–
what does this even mean, @dwarkesh_sp, “the real deal”? is it even a falsifiable conjecture? what’s the evidence?
— Gary Marcus (@GaryMarcus) 9 juin 2026
and if you can agree that “If you ask an LLM a question it can't answer, sometimes it will just try to imitate reasoning without doing it”, why acknowledge the… https://t.co/4DtoGvFCCJwhat does this even mean, @dwarkesh_sp
, “the real deal”? is it even a falsifiable conjecture? what’s the evidence? and if you can agree that “If you ask an LLM a question it can't answer, sometimes it will just try to imitate reasoning without doing it”, why acknowledge the -
Suspicion that Anthropic’s safety measures are performative
By
–
Starting to suspect that Anthropic's putative security and safety considerations are largely posturing and performative.
-
Claude-Fable-5 refuses a third of questions on BullshitBench
By
–
Claude-Fable-5 on BullshitBench: it refused a THIRD of questions (with multiple attempts). Excluding refusals, it performed well, worse than some other Claude models (chart below).
-
Anthropic protects its own IP while using others’ IP freely
By
–
Anthropic didn’t just add guardrails to make Mythos safer; they added guardrails to protect their own IP. *Their own IP*. They are still as happy as fuck to build their AI on other people’s IP.
-
Anthropic’s guardrail concerns versus model unusability regrets
By
–
I understand that Anthropic's concerns about the model being misused without guardrails are significant. And I take that seriously. We're talking about a technology with unforeseen potential. However, the fact that it was, in some cases, literally unusable is regrettable.
-
LLM shadow-banning was not planned for 2026
By
–
The shadow-banning of LLM development was certainly not on my bucket list for 2026.
-
Anthropic’s new Fable 5 safeguards quietly limit effectiveness
By
–
Anthropic’s new Fable 5 safeguards are fascinating. When the model is used for frontier LLM development, it apparently does not simply refuse or warn the user. Instead, it quietly limits its own effectiveness through techniques like prompt modification, steering vectors, and
-
Anthropic would limit capabilities to maintain competitive advantage
By
–
Pretty crazy this that is being shared where Anthropic would be limiting the model's capabilities when used to improve and create better LLMs. They sell it as a security measure but it is clear that they do it to maintain their competitive advantage.
-
Anthropic’s strict guardrails block simple questions until June 22
By
–
The guardrails are way too strict. Even the simplest questions get cut off immediately. And it's only on the schedule until June 22nd. Damn, Anthropic really thinks the model is too powerful.