🐸 Take 1: Time for European AI
A wave of European-native legal AI startups were in the news this quarter: Belgium's Duely (M&A AI, raised €1.1M), Germany's Bayshore (regulatory compliance AI, raised €6.9M) and Italy's Lexroom (Series B raised €42.9M).
Is a European-built tool with local data residency worth a capability trade-off against the US incumbents, or not yet?
🐸 Take 2: Spryngbase Legal Speak is here
Our new series showcasing stories of legal professionals using AI is here.
🐸 Take 3: The Fable of Claude
On 9 June, Anthropic released Claude Fable 5, its first "Mythos-class" model for general use, posting 80.3% on SWE-Bench Pro, ahead of GPT-5.5 (58.6%) and Anthropic's own Opus 4.8 (69.2%). Days later, the US ordered Anthropic to suspend Fable 5 and Mythos 5 access for foreign nationals. As an EU professional, you can't use it unless that order lifts.
What does this mean for EU lawyers? Dependency. A tool you build a workflow around can be switched off by a government you do not vote in. If your core AI tool vanished tomorrow, what is your fallback?
🐸 Take 4: Kirkland is playing the data game. Are you?
Artificial Lawyer reports Kirkland & Ellis's $500M technology push may include fine-tuning open-source models into its own legal AI, rather than renting capability from vendors.
No small or mid-size firm can match that budget, and you do not need to. The signal is what a firm that size is paying for: control of its own data and templates as a strategic asset. You can act on the same instinct at a fraction of the cost. The first step costs nothing: ask your vendor exactly where your client data sits, and get the answer in writing.
🐸 Take 5: Four US platforms, one room — Palantir joins
Artificial Lawyer reports Palantir has entered legal tech through a private-equity platform alongside Kirkland & Ellis, making it the fourth US hyperscaler in the room with OpenAI, Anthropic and Microsoft.
For a small EU firm, this matters less for the Kirkland deal and more for what sits underneath the tools you are evaluating: increasingly, one of four American platforms. Worth raising with peers, when the vendor is really a hyperscaler in a trench coat, where does your client data actually live?
🐸 Take 6: AI For EU Lawyers in the next 6 months
A few months ago we compiled a research report asking Belgian lawyers how they saw AI impact their work in the coming 6 months. Now we are doing this research for EU lawyers. If you would like to participate feel free to fill in the form via the link.
🐸 Take 7: The label rules
On 10 June the Commission published its final (voluntary) Code of Practice for marking and labelling AI-generated content. The detail that matters: the Article 50 transparency duties still apply from 2 August 2026. The Omnibus delayed the high-risk rules, not these.
Content under genuine human review is exempt, which covers most firm output, but client-facing chatbots and AI "public-interest" text are not. So the practical question is not whether you label everything, it is whether you can show review happened. How are you documenting human review today?
Spryngbase is a community for AI forward legal professionals in the EU. We are looking for our first 100 members…