Highlights
Asia opens the AI/art desk.
Kicking off her segment, Asia (@byasiaminor)* covered Midjourney V8.2's push toward realism and personalization, and made the case that ChatGPT's image generator now "kills it" on text — while noting she never takes raw output straight out, using it as inspiration to build a real comp in Photoshop.
The galleries aren't opposed — they follow the money.
Citing an Artsy survey of 300+ gallery pros, Asia noted most gallery artists aren't using AI to create, but 60% of galleries report artists using it for ideation and rendering — much like Diego Rivera photographing scenes to sketch his murals.
"As old as art history itself."
Flick (@0xFlick)* drew the synth-and-drum-machine parallel — music once dismissed, now beloved — and expanded it: every groundbreaking medium, from photography to digital tablets, meets traditionalist pushback before it's absorbed.
The Anthropic CEO bit.
The room lit up over a report that Anthropic's Dario Amodei worried new hires joined "for the pay." Asia's incredulous riff — do these people not need to eat dinner, and why is he fretting about payroll instead of "creating something that doesn't kill us all" — was the comic peak of the AI news block, capped with a vow to never pay for Claude out of spite.
Fame Society marketplace preview.
Flick's big build update: an immutable marketplace tied to the existing FAME contracts, in final testing. Stake a Fame Lady NFT and earn a share of purchase fees; unrevealed NFTs become buyable at a ~5% premium (1,050,000 FAME) payable in FAME, USDC or ETH; and you can finally pick exactly which FAME token to liquidate without the old "side wallet" dance. His summary: "Fame is money." 26:30
The Cold Card drain.
Flick's centerpiece — a clear, patient explanation of how a firmware update accidentally swapped a hardware RNG for Python's insecure random generator, shrinking the seed space until attackers could brute-force addresses and drain ~600 BTC in two blocks. A genuinely sad event, he stressed, hitting people who'd stacked for years and done nothing wrong. He then noted the flip side: anyone who used a Cold Card passphrase may be able to use the same exploit to recover their own coins. 45:00
Chinese models as the only defense.
When Bitcoin's technical leaders needed to analyze the hack, US labs refused on guardrail grounds, so the new AI Red Team Bitcoin turned to Chinese open-weight models — now scanning the whole Bitcoin codebase and reporting vulnerabilities. Flick's takeaway: "China knows what's up." 41:16
"This is China — they ripped it all off."
On MiniMax H3, trained on a mountain of US copyrighted characters, Asia deadpanned that they "must have had their licensing papers in order" — Flick's laughing rebuttal landed the segment's second big laugh, followed by predictions of incoming American lawsuits.
Why we can't have nice things.
Flick's tour of the 24-hour Google Earth × nano banana 2 experiment — pulled after users mostly generated destruction imagery — and the pointless futility of removing it when anyone can just screenshot and paste into nano banana anyway.
The community close.
Jilly (@JillyRappaport)* delivered the emotional finish: the vault at 106 Ladies en route to 200, a fresh 5-Lady donation from Jeremy, council-member versions from the Honeys/FLSoc Seasons collections headed to the vault, and OG artist Nex donating his "fired up" music token — meaning all his future music flows to the vault. Her frame: a community-owned project where "no one owns anything… this is us."
Topic timeline
| Time | Topic |
|---|---|
| 3:28–4:12 | Welcome, Ep. 119 intro |
| 4:12–13:33 | Midjourney/ChatGPT, Artsy gallery survey, new tools debate, Oscar AI rules |
| 13:33–15:26 | Anthropic CEO "hired for pay" riff |
| 15:26–21:47 | Spotify AI remix/Merlin deal, book-scanning legality, Apple × Klarna leasing |
| 23:00–28:05 | Fame Society marketplace preview |
| 28:05–33:41 | Open-weights politics, US vs. China AI, the "pause" letters |
| 33:41–47:37 | Cold Card wallet drain, self-custody safety, recovery via passphrase |
| 47:37–53:52 | MiniMax H3, DeepSeek V4, Google Earth nano banana 2 experiment |
| 54:04–57:00 | Robinhood chain, tokenized stocks paired to meme coins |
| 57:59–1:02:21 | Vault/community update, donations, wrap |
Notable quotes
Art in a sense is super easy. You keep creating until it makes you feel something, and then it'll make someone else feel something.
My concern is not fast… my concern is quality.
The resistance to new tools is as old as art history itself.
This is China, this is China. Nah, they ripped it all off.
This is why we can't have nice things.
This is a community-owned project. No one runs anything, no one owns anything. This is us.
(Timestamp omitted where the transcript didn't carry an exact stamp for the line.)
Who said what
- Asia (@byasiaminor)* — opened the show and ran the AI/art news desk (Midjourney, ChatGPT, gallery survey, Oscars, Spotify, Apple), arguing AI is one more color on the brush but no substitute for doing the work.
- Flick the dev (@0xFlick)* — the technical anchor: previewed the Fame Society marketplace, dissected the Cold Card drain and Bitcoin security, and toured the new Chinese models, Google Earth's short-lived experiment, and Robinhood's tokenized stocks.
- Jilly (@JillyRappaport)* — closed with the community/vault report: donations, council-member collectibles, and Nex's music token, reinforcing the volunteer, community-owned ethos.
- FameLadySociety WE (@FameLadySociety)* — host, bookending the room with the intro and the "see you next week" sign-off.
Worth a full listen
- 33:41–47:37 Flick's Cold Card explainer is the episode's standout: a genuinely clear walk from "how seed phrases stay secure" to how one bad compile drained ~600 BTC, and the surprising passphrase-recovery twist. Summaries flatten the logic; hearing it built step by step is the payoff.
- 13:33–15:26 The Anthropic-CEO riff, better in Asia's own rising-incredulity delivery than on paper.
- 57:59–1:01:44 Jilly's community close, the emotional heart of the room, with the vault milestones and the Nex donation landing in full.
* some voices are identified from context; those names are marked as likely.
