Dranneve

[Derya Unutmaz, MD @DeryaTR_]

These are such typical deliberately misleading Dario Amodei arguments! Here is a short refutal (with the help of GPT-5.6 pro) of this intellectually dishonest letter. I’ll post much more detailed long version later and will be talking with @peterottsjo tomorrow to breakdown his biosecurity arguments.

Anthropic’s essay is not a defense of open weights but a carefully worded attempt to preserve the appearance of openness while constructing a regulatory moat around frontier AI: it praises open models only when they are weak enough not to threaten proprietary incumbents, then invokes speculative “nightmare scenarios” to justify chip embargoes, anti-distillation rules, mandatory testing, centralized monitoring, and government-enforced barriers that disproportionately burden open developers, startups, researchers, and foreign competitors while protecting Anthropic’s own API business.

Its core premises are either unsupported or internally contradictory:

> a secret military model would ignore global testing rules;

>China’s AI progress cannot be reduced to access to American chips because algorithms, domestic hardware, engineering efficiency, and specialization also matter;

>distillation is a standard industry technique, not inherently theft or a national-security offense;

> and current safety evaluations are too immature, gameable, and context-dependent to function as reliable release licenses.

The essay also counts every possible misuse enabled by open access while largely ignoring its benefits, independent auditing, privacy, scientific reproducibility, local control, defensive cybersecurity, resilience, competition, and freedom from corporate censorship or withdrawal.

Most tellingly, Anthropic asks the public to accept that policies which restrict rivals, preserve closed-model control, raise compliance costs, and protect its commercial lead are motivated only by safety; whether or not that is Amodei’s sincere belief, the practical effect is unmistakable: concentrate frontier intelligence in a handful of powerful corporations and let those same corporations help decide who else is allowed to build, study, or release it.

[David Sacks @DavidSacks]

Anthropic maintains that it is entitled to train for free on all the world’s output, even if the author objects. But if a competitor trains on Anthropic’s output after paying for it, that is IP theft. The hypocrisy is breathtaking.

[Dranneve @dranneve]

@IntCyberDigest @MatthewBerman

[International Cyber Digest @IntCyberDigest]

‼️ BREAKING: Anthropic has embedded hidden spyware-like code in Claude Code that covertly targets Chinese users. It then sends information regarding every user by injecting it into their prompt message.

Claude Code is sending info like timezone, proxy and possible AI Lab connections into the system prompt in ways Chinese users can't notice.

A coding agent with repo and command permissions should not silently hide routing metadata inside prompts. This is a serious breach of user trust.

[Mark Kretschmann @mark_k]

Anthropic was found sabotaging users again, this time inserting spyware into prompts.

Can this company even be trusted anymore?

[Dranneve @dranneve]

@tbzukyu Yes if the so called momentum of risa gets derailed, it's not on bam, it's on risa. She needs to change her tune. She needs to show she's willing to admit that there's a better stance.

[Dranneve @dranneve]

@4theRepublicPH Bam doesn't need to walk on eggshells, Risa is the one in the wrong, she should course correct about her knee jerk reactions about gaming. This is not a cult, if she's wrong she should be willing to admit.

[Haider. @haider1]

this guy single-handedly damaged the AI progress we made over the last three years and pushed us toward a dystopian future

thank you for your fear-mongering services

[Matthew Berman @MatthewBerman]

Anthropic just struck a deal with the government to allow 100 select companies and governmental agencies use Mythos.

The government and anthropic are now deciding who uses frontier intelligence. Hopefully this is just Mythos and not the standard for all frontier models going forward.

[Matthew Berman @MatthewBerman]

Dario ruined it for everyone

[Matthew Berman @MatthewBerman]

Limited preview - Dario ruined it for all of us.

[Dranneve @dranneve]

@maximilian_ It's actually real.

[Taelin @VictorTaelin]

great fucking job, Anthropic

incredible fear-mongering

fuck progress, fuck science, fuck technology

fuck the whole world except for US

let's all go to the stone age together

[Dranneve @dranneve]

@AnthropicAI This is what you get for so much fear mongering.

[Karen Davila @iamkarendavila]

Senators were seen crying yesterday as a colleague was being arrested for plunder.

This is in no way an indictment of the accused senator but I would like to remind our lawmakers…

Iyakan n’yo ang ninanakawang Pilipino.
Iyakan n’yo ang binabaha dahil sa nakawan sa flood control.
Iyakan n’yo ang gutom nating mga kababayan.

Get out of your bubble. Please.

[Dranneve @dranneve]

@OfficialLoganK Disappointment. The new quotas are awful and penalize people for engaging in long chats. The 3.5 flash model is also basically a pro model labelled as flash.

[Dranneve @dranneve]

@mamamwoah @SecLazaro Do we really want to go to the chinese bloc?

[Kyle @kavickers23]

@hasanthehun