So everybody is pumping out these cool websites and video games that Fable can make in one prompt (no shade). If you guys look through any little bit of the system card for this thing, it's fucking terrifying. THOUGHTS!? https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c342ee809620.pdf 1. Silent degradation for AI/ML work This is the biggest one. Anthropic says Fable 5 has special safeguards for frontier LLM development. But unlike cyber/bio/chem safeguards, these are not visible to the user. “Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user.” They say Fable 5 will not fall back to another model. Instead, they may reduce effectiveness through: - prompt modification - steering vectors - PEFT/fine-tuning interventions The user still gets an answer, but it may be an altered / steered answer without disclosure. You don’t know if Claude is failing because the task is hard, because your prompt sucks, or because Anthropic quietly nerfed the model. 2. Anti-competitive access split: Mythos for trusted partners, Fable for everyone else Anthropic frames this as responsible deployment: - Mythos 5 = stronger model, trusted partners only. - Fable 5 = public model, safeguards added. The system card says Mythos is available only to a small set of vetted/trusted partners, starting with Project Glasswing. So we've established precedent for intelligence class divide. The most capable model goes to governments, big labs, banks, and major tech companies. Everyone else gets the restricted. 3. “Safety” classifiers can become capability control For cyber, bio, chemistry, and distillation, Fable may route the user to a weaker model or block the request. Anthropic presents this as necessary risk reduction. Fair enough for bioweapons/cybercrime. But the category list includes distillation attempts and frontier AI

It's March 16th, and I love wrestling. Happy 3:16! A couple of things over the weekend really hammered home the importance of flexibility. Algorithms are changing. Tools are changing. Everybody is pissed about Antigravity's new usage limits right now. It just burns in the importance of owning your shit. Data, stack, audience, can't trust things will stay the same. What's something you're building right now to make sure you actually own your setup? And that’s the bottom line. 🖕🖕

The first month I could cover my bills with online money came after a moment of total defeat. I’d been spinning my tires for months, landing a few piecemeal clients on upwork - my outreach efforts were totally wasted. I'd signed exactly ZERO people through cold email, and couldn't close anyone in "my niche" to save my life. Money still going out, nothing coming in ... completely stalled, feeling defeated. I needed to change or quit — so I came up with a plan. I stopped trying to convince strangers. I took a step back, a deep breath, and I came up with a solution to a problem that I PERSONALLY would pay to have solved, and selling it to people who already trusted that I get it. Context! I was deep in the remote music producer world and T H E problem we all had was consistent clients, not just word-of-mouth. So I solved that problem! I took my new AI skills and scraped Bandcamp / IG / X and built lead lists of bands giving ready-to-work signals: - posting about writing - teasing a new song - “headed to the studio” - hiring / looking for help / asking for recs Then I sold those lists to producers I already knew in Discord communities… because it was literally the only thing we all talked about. That month I sold 11 lists at $500 USD each. It saved people a full day (or more) of deep research, and it fit the budget/value. TL;DR (at the end haha) - start with a problem you actually understand - sell to people who already live in that problem - package one pain into one deliverable - build the snowball, and don’t compare your level 1 to someone’s level 9 If you’re stuck, don’t start with “an offer.” Start with a pain you understand. 🤔What’s ONE problem you’d pay someone else to solve for you right now? In my case, it was: “Find bands that are ready to work.” What’s your version of that? Drop it below. I’ll reply with 2–3 ways you could package that!

With all these AI coding assistants dropping, curious to see what the community is actually building with. Drop a comment on: - Why you chose your tool - What you're building with it - Biggest win or frustration so far Let's see what's actually getting used vs what's just hype.

Okay, Ralph Wiggum - a plugin for claude code is an iteration loop that keeps the agent from quitting a task until it's "DONE"... Read about this: https://awesomeclaude.ai/ralph-wiggum kinda fuckin' love it. not free reign or proceed "dangerously" - keep answering my annoying questions until the mission is accomplished or we all die trying haha. You see a use for this in your day-day?