FLSoc Presents: "Fame Ladies on Top" Ep. 126 — licensing wars, the Magic Eden exploit, and half-price compute

Hosted by FameLadySociety 💋 WE (@FameLadySociety) with cohost Jilly Snipes · Tue, Sep 29 · 52 min · 4 voices (Flick the dev, Byasiaminor, Jilly Snipes, plus the intro/outro music).

TL;DR

• Asia (@byasiaminor)^* delivered an AI/art-business brief: nano banana image gen is now inside the Google search box, YouTube is adding Gemini draft feedback and dynamic thumbnails, Adobe closed its Topaz Labs deal, and Runway joined OpenAI's enterprise marketplace.
• The headline concept of the night: the display license. Getty × OpenAI lets Getty images show up inside ChatGPT results without being a training license — and possibly templates a thin royalty path for independent artists.
• Flick the dev (@0xFlick)^* broke down the NFT marketplace exploit: a vulnerability in Limit Break's payment processor (v2 and v3), where only v3 could be switched off, leaving Magic Eden-era approvals live. 0xQuit's counter-drain rescued a large chunk; a MEV bot offered 90% back.
• Action item for listeners: check NFTsAreSafu.xyz or the FameLadySociety site, or just go to revoke.cash and clear every red-alert approval, wallet by wallet — approvals are per collection and stay dangerous even for empty wallets.
• FAME was hit: Flick lost 3.2M from one wallet, and Jilly put total project losses at close to 20 million FAME.
• Also covered: Bitget's $387M credential-based theft, OpenAI's new "dots" agent plus a 50% usage cut on the $200/mo tier (and a new $500 plan), Anthropic's reported $518B in infrastructure commitments, and why 84% of Iran-linked sanctioned wallets run on USDT.

Highlights

[4:12] The AI-tools roundup, delivered like a morning brief. Asia opened episode 126 by pointing out that licensing deals and studio tools are "moving faster than court cases in America," then walked through the week: a banana icon under the Google search bar, Gemini-powered draft feedback on YouTube, the Adobe/Topaz close, and Runway landing in OpenAI's corporate marketplace alongside 11 Labs and GPT images.

[7:11] Getty × OpenAI, and the birth of the "display license." Asia's central point: this is not a training license. It only covers displaying Getty images inside ChatGPT — a distinction she expects to become "the new keyword for the art industry, especially when you're dealing with museums." Her worry is the split: Getty collects the OpenAI fee, the contributor split and permission rates, which could leave independents with pennies.

[10:01] Web3 showing up for itself. Turning to the Magic Eden hack fallout, Asia described the community response — people buying NFTs for each other, 0xQuit spending days and his own money — as something she'd never seen in corporate life. "The blame game is not the frequency we ever want to be on," she said, before giving a shout to Quit and his team.

[12:30] Flick with the receipts. Flick added the detail: a MEV bot that had picked up open WETH offers published on Magic Eden offered a 90% payback, keeping ~5% plus gas, and that return is already wired up at NFTsAreSafu.xyz for wallet-connect checks.

[14:31] Jilly on the culture's unwritten rule. Jilly framed it plainly: a hacked or stolen asset is stolen property, and stolen property gets returned. Some people just hand it back, some ask for their cost — either way, she said, honoring that has always been "a huge part of our culture," and she was glad to see it still holding.

[28:33] The exploit, explained properly. Flick's technical walkthrough: it kicked off September 24 with ~240 ETH of NFTs slurped up; the bug lived in Limit Break's payment processor, used by both OpenSea and Magic Eden; Limit Break could turn off v3 but not the older v2 that Magic Eden components were built around. Then [30:15] the practical part — revoke.cash, hunt the red alerts, one revoke per collection — with the reminder that an approval is dangerous even on an empty wallet, because a future transfer in can still be drained.

[32:58] The FAME damage, including one genuinely galling flourish. Flick found a still-vulnerable approval two and a half days after the exploit, and noted the 3.2M FAME lost from another wallet came via Society NFTs — whoever did it topped up 888k FAME to mint another Society NFT before taking the lot. Jilly's verdict was one word, twice: "That was so gross." She added that close to 20 million FAME has been stolen from the project lately, which is the unfortunate reason it's currently cheap.

[41:17] The bill comes due on cheap compute. Flick, self-described pro max user, on OpenAI's new pricing: a 50% cut in effective plan usage on the $200/mo tier, with a $500 tier now available to buy the difference back. He's been saying on this show for a while that these subscriptions were too good a deal — and gave the honest read that they're now "half as good as they used to be. Still pretty pretty good." The lower $20 and $100 plans are untouched.

[43:54] Anthropic's non-cancelable future. From leaked IPO prospectus docs via Reuters: at least $518B of infrastructure spend over a decade, roughly 80% payable regardless of usage, and a ~$42B 2025 net loss that includes about $34B in non-cash financing charges — so not all cash burn. Flick's nuance: if a big efficiency breakthrough lands, you could be contractually locked into buying compute that's suddenly inefficient.

[46:28] Tether, Iran, and an accidental Zcash thesis. A Senate Democrat staff report found 84% of 846 Iran-linked sanctioned wallets used USDT; Tether claims to have frozen nearly $550M this year. Jilly's dry follow-up — "I'm surprised they don't like Zcash" — got Flick to concede that might explain the pump, followed by the eternal request to be told about alt runs before they run.

Topic timeline

Notable quotes

• "Display license is going to be the new keyword for the art industry, especially when you're dealing with museums." — Asia [10:01]
  ▶

  0:38
• "That's a beautiful thing about the blockchain, because there's no hiding it, you know. You know how it got there, where it is." — Jilly Snipes [16:33]
  ▶

  0:38
• "You can't right-click and save and think you're okay, because they will come after your ass in a hot ass minute." — Asia [22:19]
  ▶

  0:38
• "First off, senior vice president of AI sounds like an amazing role. And I think something I aspire to." — Flick the dev [27:04]
  ▶

  0:38
• "This is open claw for the masses." — Flick the dev [40:06]
  ▶

  0:38
• "I'm hearing pay another $300 for 50% more usage, or we're cutting you in half." — Flick the dev [41:17]
  ▶

  0:38

Who said what

• Asia (@byasiaminor)^* — ran the AI-and-art-business segment: licensing structures, contributor royalty funds, AI clauses in her own freelance contracts, and a heartfelt read on how the community handled the exploit.
• Flick the dev (@0xFlick)^* — opened the room and carried the technical load: exploit forensics, revoke instructions, Bitget, OpenAI releases and pricing, Anthropic's IPO numbers, stablecoin freezes. Also the show's resident aspirant to a senior VP of AI title.
• Jilly Snipes (@JillyRappaport) — cohost; the culture-and-ethics voice on stolen assets, FAME's losses and current price, plus the seasonal report that fall melancholy is bad for the mood and good for the art.
• FameLadySociety WE (@FameLadySociety) — handled the musical bookends, including the closing declaration that it is, in fact, a rap.

Worth a full listen

• [28:33]–[32:58] — Flick's exploit walkthrough and the step-by-step revoke instructions. This is the part with actual homework in it, and the per-collection/empty-wallet nuance is easy to miss in summary.
• [7:11]–[10:01] — Asia's full unpacking of the Getty deal, including the fee splits and why the earlier Getty v. Stability "model equals infringing copy" theory fell apart in the UK.
• [49:37]–[50:28] — Short, and the closest thing to a running bit: Zcash, an alt-coin pump, and Flick asking to be tipped off in advance next time, desperately.

* some voices are identified from context; those names are marked as likely.