Mehdi Ataei @AtaeiMe

@Ethan_is_online GPT models have been like this since 5.3. I think it’s the result of heavy RL. Fable is a bit better still.

@waterloo_intern Nice joke, but polluting huggingface is not cool

Exactly. I've been disseminating a similar message for years. The concentration of power in AI and the desire for control is by far the biggest danger of AI. It could lead to a few private companies and/or countries being in control of access to information, access to knowledge, and access to the tools of economic expansion. It's a kind of medieval obscurantism akin to the Ottoman empire banning the use of the printing press for 200 years, in part to keep control of the dogma, but also to protect the corporation of the calligraphers and scribes. Relevant historical bits about the Internet: 1. It took a deliberate decision by Al Gore and Bill Clinton to open up access of what was then ARPAnet to commercial entities and to the public, against the desires of the entrenched telecom industry. During a public roundtable about the "information superhighway" in 1993, the CEO of AT&T told Gore and Clinton "leave it to us". Gore said no. 2. In the late 1980s, setting up an Internet presence required buying proprietary hardware with proprietary OS and software stack from Sun Microsystems, HP, IBM, or Dell. By the 2000s, all of this was wiped out by commodity hardware, Linux, Apache, and an entirely free/open software stack. This migration to open platforms was the result of market forces. Infrastructure wants to be open. Foundation models are becoming an infrastructure and will inevitably become commoditized. Long term, the money is in the application layer, which is what I, Arthur Mensch, Alex Karp, and others have been saying.

Be extreme, Canada.  That is not an ideological plea, but an appeal to a forgotten inheritance and a hope that we find it within ourselves to be extreme, again. It is a recognition that to be Canadian is to hold two opposing truths in your mind at once: Canadians are mild and moderate but Canada itself is harsh and brutal. F. Scott Fitzgerald once wrote, the test of a first-rate intelligence is the ability to hold two opposed ideas at the same time and still retain the ability to function. Canada is that test, made manifest. Canada is not gentle. It is not mild. It is not moderate. It is extreme in scale. Extreme in climate. Extreme in beauty. Extreme in wealth—of resources, of land, of possibility. It is a country that demands resilience from those who inhabit it, whether they admit it or not. And so we arrive at the paradox: a people perceived as restrained, living within a landscape that is anything but. This is not a contradiction. This is the source of Canadian genius. Because what emerges from this tension is not weakness. It is control. It is discipline. It is a quiet form of ambition that does not need to announce itself to be real. It’s breaking your leg, helping your team score 6 goals and waving to the crowds as you’re carted out on a stretcher. To be a Canadian extremist, then, is to embrace both halves fully. It is to reject the caricature of smallness while also rejecting the need for bluster. It is to understand that politeness is not passivity, and that restraint is not the absence of ambition, but often its most disciplined expression. It is to build in a way that reflects the land itself: vast in scope, resilient in structure, and uncompromising in its standards. Canada does not need to become something else to compete. It does not need to abandon its character to achieve scale. The path forward is not to resolve the contradiction, but to lean into it—to recognize that the tension between humility and enormity is precisely what makes this country capable of extraordinary things. The nations that shape history are often the ones that understand their extremes. Canada’s mistake has been to ignore its own. I am a Canadian extremist because I refuse to.

@petergostev Really hope OpenAI improves the model writing. It’s really hard to the writing of GPT models

I’m really glad that @DavidSacks is working closely with the US government. He's one of those rare people who reads between the lines and doesn’t get fooled by empty rhetoric.

David Sacks @DavidSacks

2 days ago

Legacy Media types are calling this Alex Karp interview a “crash-out” so that’s your first clue that he is actually saying something extremely insightful. He is articulating what real “AI safety” looks like in the enterprise. Not abstract alignment research or certification by a

643 2K 13K 1.9M 5K

Inevitable future

Julien Chaumond @julien_c

2 days ago

if even palantir advocates for open models now...

17 13 178 16K 27

You could argue that Linux, electricity, and monitors also contributed to the attack (the attack wouldn’t have been possible without the use of a monitor) so the evidence is irrelevant. Nobody is saying coding models can’t enable cyber offense but if that’s blocked it will only hurt defenders. Given equal access to model capabilities defensive teams will always win, by a huge margin.

@alexstamos Many security researchers are already on PRC models, unfortunately

@adityaag Crazier that China is the leader of freedom and access. How bad we fumbled this 🤦

@teortaxesTex People were saying similar things about GPT-4, and here we are.

@Sentdex The average public must become more aware of the danger of AI only being in the hands of 4 people. Maybe then they actually support open source instead of just hating AI

@joshmanders @bcherny Same issue with all LLMs. They assumed the person reading / using has the same context as them.

Thanks for nothing.

Sridhar Vembu @svembu

4 days ago

Databricks is embracing Chinese open source model GLM-5.2. Earlier, Microsoft said they were looking to embrace DeepSeek. Major American software companies are embracing Chinese open source models to cut token costs. I have long advocated using the Chinese models running in

89 321 3K 231K 390

@vishnuvig @ClementDelangue Many, many such cases recently. Crazy that this is actually happening

Two friends who work on cybersecurity have completely switched to open models (GLM) because they got sick of being flagged by closed models. What exactly did you accomplish? The attackers never needed your permission. The defenders did, and now they're gone.

@ZixuanLi_ Managing subagents

@KavitaAlguMD Yeah wth was that question