luke
96 posts

luke
@LukeArrigoni
Working on IP/NIL in GenAI at scale through @lotiapp.
Seattle شامل ہوئے Ağustos 2010
96 فالونگ142 فالوورز
luke ری ٹویٹ کیا

Today we are launching Interchange, infrastructure for consent, control, and compensation in generative AI.
Generative AI has scaled without a way to license real people, voices, and creative works at the moment of creation. Interchange fixes that. Rights are cleared at generation. Attribution is built in. Compensation flows back to rights holders automatically.
Consented, compensated AI generation, by default.
Read the full announcement: prnewswire.com/news-releases/…
#LotiAI #Interchange #GenerativeAI #DigitalRights #NIL #ConsentInfrastructure
English


Target prioritization. If you have more data than you know what to do with and the order of operations for a strike may mean the difference between life, death and nuclear war then you’ll want to synthesize all that data down to a matrix of 900 priority targets and in what order to minimize regional losses. It’s not on the drone, it’s in the war room.
English

@ns123abc I don’t believe this.
Someone explain how you could use a LLM to command and control airstrikes.
I’ll wait.
English

🚨 HOLY FUCKING SHIT
U.S. STRIKES IN MIDDLE EAST USED ANTHROPIC’S CLAUDE
>be anthropic
>embedded in classified systems
>see what they’re building
>say “we won’t allow autonomous weapons”
>pentagon: we need this NOW
>3 days or we destroy you
>anthropic: no
>banned
>same night
>drones hit iran
>using claude to kill people
YOU CANNOT MAKE THIS UP


English
luke ری ٹویٹ کیا

Mitchell Smith just wrapped with Digital Hollywood at The Creator Economy & Unique Vision: Algos, Analytics & Monetization.
X is building a data-driven platform where creators own their audience, algorithms reward authenticity, and monetization scales fast #CES2026.

English

@yikeningmeng333 Neural networks are already this way. Parameter interactions are too combinatorially complex to trace, high-dimensional embeddings impossible to intuit. The question isn't whether we trust what we can't fully understand. We already do, constantly.
English

If a system is too complex for a human to read, it’s too complex for a human to trust. 👁️⛓️ When coding becomes 'Fly-by-wire,' the user loses the last bit of agency over the 'Off' switch. In crypto, we call this the 'Black Box' risk. We don't need AI-written stealth planes in our financial protocols; we need open-source transparency that humans can audit. Efficiency shouldn't be a Trojan horse for absolute AI control.
English
luke ری ٹویٹ کیا

Deepfakes aren’t slowing down. Neither are we.
Loti AI CEO @LukeArrigoni joined @richontech on @KTLA to discuss the surge in deepfakes, AI-driven impersonation, and how Loti gets harmful content taken down fast.
🎧 Full conversation:
omny.fm/shows/rich-on-…
English

@UpdatesFromYT @sal19 @zachvallese This is what happens when someone can tie a name to biometrics: forbes.com/sites/paultass…
@UpdatesFromYT @YouTube please tell us again how you’re not training on biometrics.
English

@sal19 @zachvallese 2/2 So why does this tool exist? To give creators the same control over their likeness that they already have over their copyright It’s optional, protective, and built to give control back to creators
English

YouTube is stuck between Google's AI ambitions & serving creators. No where is that fissure more palpable than YouTube's requirement that creators upload their gov ID and biometrics in exchange for a deepfake detection and removal tool cnbc.com/2025/12/02/you…
By @zachvallese
English

1/2 This is a false "alarm" The data creators provide for likeness detection is not used to train Google’s generative AI models. We caution against giving weight to the unfounded claims of competitors who profit from selling paid alternatives to the free safety tools we now provide It does not appear that the experts quoted in this story were advised by lawyers who reviewed our consent language
English

@UpdatesFromYT @sal19 @zachvallese Since you're talking about us, Loti isn't competitive with this program. We saw this as an excellent way for our customers to get even faster takedowns on YT. It's disappointing that you won't modify your privacy policy to say you won't train models on biometrics.
English

@Div_pradeep @higgsfield_ai The cringiest thing. I feel it in my stomach.
English

San Francisco–based AI startup Higgsfield AI (@higgsfield_ai) admitted on October 18th that it went too far by replacing work for over 100,000 artists.
To make things right, the team is giving up their November paychecks so those same artists can afford the AI tools.
Prices have been reduced by 65%. The offer remains available for 3 days
English

Set an alarm for 9a and another for 3p, go sit in your car for 45 minutes when they ring. Now send me a check for $2k/mo. This is what the first 4 years of childcare feel like. Not being able to take calls during prime times and having to pay someone to watch your kid for a few hours. Then you have the guilt/regret of missing a huge portion of their life working. It’s a career killer for the parent that stays home if you decide to not go this path, too.
English
