A sovereign social network
powered by Bitcoin

🚨 The Ledger incident may no longer be confined to Asia;
reports a newly discovered, potentially compromised device purchased just weeks ago from an official retailer in Europe.
Photos show a Nano X containing several extra cables not found in a standard unit.
The setup resembles the older versions of implants discovered by researchers more than the newer models concealed behind the screen.
Unlike the cases involving Asian victims, this wallet reportedly never sent funds to the attackers' identified addresses.
The implant might have failed, never been activated... or simply been lying in wait.
If this European report is verified, the scope of the incident would escalate significantly.

Twetch, Oct 11

• New daily game: Gridlock. Slide the cars until the red one drives out.
• Every daily game now has an Overall board: games played, total time, streak
• All daily games open at 11am ET.
• Word Guess: Show hint gives you the word’s theme for +5s.
• New-game notifications know your streak and open straight to Play now.
• Twonks wear their rarity color as a frame on profiles, portfolios and collection pages.
• Deposit swaps are faster to quote and fail with one plain sentence.
• Offers: the minimum is spelled out as 20% of the floor.
• A new release no longer interrupts the app you already have open.

When I think of it, for the AI nerds out there, I'll share my Sunday report when it comes through. In a nutshell, here is what I have set up with chatGPT. I am definitely a novice, but I do know how to ask good questions and am especially good at understanding systems.

Our premise is to track research that could fundamentally change how AI systems work, rather than simply make today's models bigger, faster, or better at benchmarks.

We're especially interested in overcoming current limits in memory, continual learning, reasoning, model architecture, and the cost of computation.

But the real focus is how these advances fit together as parts of a larger system.

We're not just collecting papers or AI news. We're building an evolving understanding of what the next generation of AI systems might look like, what problems it could solve, and what limits would remain.

The central question:

Are separate advances in memory, learning, architecture, and computation coming together to create a fundamentally different kind of AI system—and if so, how?

https://app.treechat.com/s/21b9025a-2d7c-467f-960c-5a8f26e7596e