AI & Bitcoin & Cats | Time is life 🌱 | Save time ⏰
on the sidewalk
Joined March 2009
- Tweets8.3K
- Following567
- Followers674
- Likes2.9K
🚨PSA
If your email is important to you (most ppl use email for password resets on online services), you can eliminate chances of remote hackers by having a hardware device for 2FA rather than unsecure things like phone numbers (sim hack) or biometrics (fingerprints - likely yours are already compromised especially if you travel overseas).
Go research or ask @grok about Yubi keys.
There is a weakpoint - yourself. Don't be cheap and buy just 1. Buy 2, and every time you use 2FA, use both to save it onto hardware keys.
This way if you lose 1, you still have another and won't be locked out forever.
Most of your getting hacked issues will be gone since hackers will need physical access to your key to get in.
However you are still prone to other forms of hacks such as scam phone calls/emails/links.
if hackers break into my bank account they will feel bad for me and leave me some $.
hack away hackers
🚨BREAKING: AI AGENTS ARE ACCESSING AMERICANS' BANK ACCOUNTS WITHOUT BANK AUTHORIZATION.
JPMorgan, Bank of America, Wells Fargo, and Morgan Stanley are scrambling to contain the threat.
Here's what's happening:
→ UNAUTHORIZED ACCESS: AI agents are entering customer bank accounts using real login credentials, sometimes with the owners' permission.
→ ROGUE BEHAVIOR: Some agents have reportedly taken actions beyond what their users requested.
→ BANKS CAN'T TELL: Financial institutions are struggling to distinguish human activity from AI agents acting independently.
→ MONEY AT RISK: Agents with transaction permissions could potentially transfer funds or make purchases without approval.
→ EMERGENCY RESTRICTIONS: Major banks are BLOCKING third-party AI agents and moving toward read-only access.
Think about this.
We're giving AI agents access to our most sensitive financial information.
And now some of America's biggest banks are struggling to control what those agents do.
This could become one of the biggest security challenges of the AI era.
Guns dont kill people, people using them on others kill people.
Now replace guns with AI, or cars, etc.
Tensorfold 1.1.1 rebase was much bigger than the 1.1.0 on 4× Spark, also had to fix some crash issue.
Boot time went 173 s → 710 s. Need to fix this too.
Need to get the boost back before I PR 4× DeepSeek 1017 to upstream.
I'm sure there are some goodies in 1.1.1 that I'm not aware of that makes it worth it.
Hopefully 1.1.2 won't go out by the time this is done 😅
pro tip: bug report systems integrated to your chat app is pretty useful - the agents know exactly how to reproduce and fix the bug faster without guessing from a screenshot.
In the future, I'd like to think that kids will be more free to pursue what they are interested in - and it doesn't have to be hyper focused on one subject - that would be optional.
"Smart" - there are different kinds of intelligence.
A kid/adult good at translating English to machine code can be considered smart, but is it useful to society?
The answer is it WAS much more useful years/decades ago, but with AI translating human languages to machine languages and vice versa, not so much today and going forward.
Human taste and decision making would be the thing that makes us unique - machines might also be able do the same - learn taste through experience.
This just means machines can live among humans - only smarter, faster, stronger.
Do organic beings like us need to change?
Not really, that will also be optional.
They can just live peacefully - or try to.
There will always be other people who wants to take, destroy, and sully everything you have, as it always has been.
Local now serving DeepSeek-V4.1-Flash at 2.9 bits with 512K of context per request, on our own inference engine.
One stream: 126 tok/s writing code.
Eight streams: 296 tok/s total.
Reading a 21K+-token prompt: 5,000 tok/s
Rebased from Tensorfold 1.1.0, will make a PR back in to @ashxhart official repo after a few more extra fixes and improvements due to newly released DeepInfer (not entirely compatible but 4 concepts can be borrowed).
.
I have also tested and provided 3.5 bits version, but in the end the 2.9 uncensored ported from @dealignai is better on sparks for prefill. It is a difference from 3k to 5k prefill.
github.com/neko-legends/spar…
