Lawyer. Founder. Geek. In love with European Startups and Venture Capital. Building legal AI that works instead of you. Arna & Nlaw.

Joined October 2009
54 Photos and videos
Those who say European regulation is killing companies, etc., typically never build a company anywhere in the world. Their purpose is to whine. Go build something instead. The regulation is not your obstacle.
1
25
Le Chaton Fat or Le Gros Chaton meme shows how desperately the space craves a European open, large reasoning LLM.
37
Does anyone understand what Claude's usage limits are, how much usage is included, how it is calculated, and how much excess usage costs? :) Asking for a friend.
1
111
Let me answer myself. x.com/polymarketmoney/status…

JUST IN: Anthropic has been sued for allegedly misleading customers on Claude Max usage limits.
20
General counsel, would you like your teams' legal AI software to be fully hosted in the EU and powered by open-source LLMs?
2
74
In the EU, you don't need to KYC your customers if you want to sell them LLMs. In the US you have to. Also, US have a regulation on the shape of public toilet seats. So today our LLMs are running and since apparently we never work in Europe we all went for a jog in the city center.
2
161
In case anyone is wondering, @arna_ai is not bothered by the recent @AnthropicAI ban and keeps running her blueprints.
1
5
147
Is that true @karpathy? We'd love to welcome you back in the EU and build here with you. Eu countries have so far blocked 0 LLMs. Only apple's Siri. But who cares about that...
just realized even Karpathy is on Eb-1 visa, and is not a citizen himself. Damn. Even he can't use mythos anymore.
1
2
162
Nejc Novak retweeted
just realized even Karpathy is on Eb-1 visa, and is not a citizen himself. Damn. Even he can't use mythos anymore.
47
179
5,616
305,131
Today is the day when we, all AI startups, tell the world we are model agnostic and emphasise the importance of such architecture. Incidentally, all European founders using Delaware c-corps might reconsider their legal structure as you never know when the US Government will ban you from selling your software due to national security reasons.
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…
1
1
5
157
Business Idea: develop alternatives to foreign products banned from the EU because they don't want to comply with EU laws.
2
81
I love it how some people, when a new AI model is released to the public, rush to brag on social media about how 'they got to test it for the last two weeks across numerous use cases'. They are so cool.
1
1
146
When Europeans travel or do business abroad, we respect the laws and culture of the destination. When others come to Europe, they keep telling us what a terrible place we are and how 'we regulate instead of innovate'. At least operating systems in the EU don't need to establish their users' ages.
EXCLUSIVE: Apple and the EU are telling very different stories on Siri AI, and right now, it's a standoff. Apple's press release and comments to @LelloucheNico say the DMA would force it to give rival assistants deep access to messages, files, photos, app actions, and personal context, creating privacy/security risks. I reached out to the EU Commission for comment, and they discussed it today (video below). They said nothing in the DMA blocks Siri AI. Apple can launch it if it complies, but it must give rivals "equivalent access" and can’t use iOS to favour its own AI. Apple proposed a “Trusted System Agent” and an 18-month rollout to ensure only approved third-party agents could run. The Commission says they weren't "DMA-compliant interoperability proposals." Whatever that means. This is a standoff over who controls the AI layer on iPhone. The EU isn’t protecting user choice here. It’s trying to regulate the iPhone’s AI layer, and EU users are the ones losing out. iPhone users can already download ChatGPT, Claude, Gemini or Perplexity. Apps like @poppysimplified can use your Mac to get system-level access for AI assistant apps. The EU is demanding default, system-level access for any third-party app that claims to be an AI assistant. Which is dangerous, especially given Apple's privacy-focused AI stance. Apple, respectably, is not bowing down.
1
1
134
The regulation requiring OS to establish a user's age makes no sense. Funny, this is happening in the USA. EU is really a terrible place for tech and doing business in general.
97
I wrote a blog about the back-and-forth between in-house lawyers and other employees in companies. I called it the 'in-house legal ping-pong'. There was not much interest in it back then, but the last four GCs I spoke with described exactly that to me.
1
2
94
Get a peace of digital signing.
Signing PDFs with certificates is a pain. Some folks still want certificate signed PDFs rather than docusign/etc. Only decent option is basically the Adobe bloatware. Or a bunch of CLI scripts. I vibecoded a native macos app real quick that does it. Dead simple but works. You draw a square on the pdf, pick your certificate from the keychain and click sign. Done. github.com/jernejstrasner/pd…
1
115
Why is 95% of enterprise GenAI showing no ROI? Not because the models are bad. Because companies bought AI as a productivity tool for individuals, then waited for the org-level result to appear on its own. It doesn't.
1
1
111
Individual productivity is the easy half. Anyone can buy it. Organisational transformation is the hard half. It requires: - Codifying the policies and decisions that today live in people's heads - Building systems that run those policies without a human in the loop - Governance that scales with adoption, not against it That is real work. Nobody's licence agreement does it for you.
1
3
83
The companies who actually move the needle don't ask "how do we make our people faster?" They ask "which work in this organisation doesn't need a person at all — and what would it take to route it directly to AI, end to end?" That question is where the next 18 months of real ROI lives.
26