LLM censoring is a big problem.
I simply wanted Grok's help to raise the speed limiter on my ebike from 25km/h. I only ride on private roads and the jailbreak would be completely legal, but Grok refuses because it considers it "unethical".
As everybody soon relies on AI for information, big corportions can easily affect public consensus and alter the future for their good.
Where is the open source non-censured LLM?
Imagine: just a couple years ago we wrote code by hand… and called it a productive day 😅
Vibes below
Working on a forestry data acquisition and analysis software, powered by ESA Sentinel-2 satellite imagery.
Timber & carbon value estimation, optimal estimated harvest date, disease identification, more.
Need test users. Let me know if you are interested
Keräsin kaikki Nasdaq Helsingin insidertransaktiot reaaliaikaiseksi dashboardiksi. Netto sisäpiirivirta viimeiset 30 päivää: -223 270 637 €. Johtajat myyvät.
I did this while having a baby on my chest, thanks @AnthropicAI
Weekend project: Aggregated all insider transactions from Nasdaq Helsinki into a real-time dashboard. Net insider flow last 30 days: -€223,270,637. Executives are selling. A lot.
Opus 4.6 is training my ml model --> analysing results and model performance --> writing down issues and improvement ideas to a local file --> improves the feature engineering and ml training scripts. And iterates on this loop until key requirements are met.
Been doing that for hours
This is insane
For most of 2024, Bitcoin tracked the 3-month-lagged Global M2 liquidity pretty closely—the thesis being that Bitcoin follows money supply expansion with about a 3-month delay.
But what's weird now is the gap that's opened up since late 2025. Global M2 (green) has stayed elevated and even ticked higher, while Bitcoin (red) has fallen sharply from its highs above $100K down to around $78K.
This is a notable divergence.
Two ways people typically interpret this:
> Bullish read: If you believe the correlation holds, Bitcoin is "undervalued" relative to liquidity conditions and should eventually catch up to where the green line suggests it "should" be. This view would say it's a buying opportunity.
> Sceptical read: Correlations break. Maybe the M2-Bitcoin relationship was overfitted to a specific period, or other forces (regulatory concerns, macro fear, risk-off sentiment, leverage unwinding) are now dominating price action in ways that liquidity alone doesn't explain.
What's your take on it?
What gets me is the universal joint. It's not just a motor — it's a complete drivetrain system with a flexible coupling that lets the propeller thrust in any direction. This thing has better engineering tolerances than most things we manufacture, and it self-assembles from genetic instructions.
This is a bacterial flagellar motor … one of nature’s most sophisticated molecular machines - a rotary motor embedded in the cell membrane consisting of ~25-30 protein types totaling 20,000-50,000 atoms in a multi-part structure of rotor, stator, drive shaft, and propeller. Operating at 100-300 Hz (6,000-18,000 RPM) with some species reaching 1,700 Hz (~100,000 RPM) at nearly 100% efficiency, this nanoscale engine features rotor proteins FliG/FliM/FliN forming the C-ring, MotA/MotB stator complexes, FlgG rod protein drive shaft, FliC flagellin propeller filament, and FlgE hook proteins acting as a universal joint, all powered by proton or sodium flow across the membrane generating torque through conformational changes in the stator complexes.
everyone's watching nvidia for the AI trade
meanwhile NRG Energy is up 80% this year. almost double nvda.
the math is simple:
> data centers hit 4% of US electricity in 2024
projected 12% by 2028
> dominion forecasts 75% demand increase from data centers alone
> gas turbines sold out through end of decade
> $7 trillion in AI data center capex by 2030
the bottleneck soon isn't chips anymore. it's electrons.
utility stocks just had their best two-year run since 2003-2004. most people still think they're "boring defensive plays"
clawdbot/moltbot/openclaw situation is wild
> 85k github stars in a week
> hundreds of instances found exposed on the public internet
> infostealers already targeting it
> researcher got it to leak a crypto private key via prompt injection in an email
>cisco found 26% of community skills have vulnerabilities
giving an AI agent shell access + persistent memory + messaging app integrations without proper sandboxing
what could go wrong
but what could go right?














