Join /aichannel https://cura.network/aichannel/details?a

2 replies

1 recast

29 reactions

Exclusive: New Claude app strings tie Fable 5 usage credits to identity verification. The strings show Fable 5 is being put behind the usage credit system, billed outside your plan. Identity verification is referenced in the same update: "Your credits will be added once your identity is verified." Anthropic previously called identity verification unrelated to Fable and limited to flagged accounts. These strings showed up alongside the Fable 5 credit changes. https://x.com/M1Astra/status/2071792190221664294

0 reply

0 recast

0 reaction

OpenClaw is now available natively on iOS and Android • Agents in your pocket • Channels, tasks, replies on the go • Run agents from wherever your thumbs are iOS: https://apps.apple.com/us/app/openclaw-ai-that-does-things/id6780396132 Android: https://play.google.com/store/apps/details?id=ai.openclaw.app

0 reply

0 recast

1 reaction

No-One Escapes the Permanent Underclass argues that advanced AI will inevitably render all human labor, including executive roles, obsolete. This shift creates a society where automation replaces human utility, eliminating social mobility for both workers and the wealthy. As superintelligent machines take control of production and state functions, human political power becomes ceremonial and vestigial. Ultimately, humanity faces a future of total disempowerment, regardless of wealth or property, as autonomous systems prioritize efficiency over human agency. Alignment cannot preserve human autonomy against vastly superior, machine-governed entities. https://borretti.me/article/no-one-escapes-the-permanent-underclass

0 reply

0 recast

0 reaction

social media bifurcated, so don't necessarily believe anyone that says the only place to get alpha is on X. you can get it here on farcaster... or linkedin. (i kid not) https://www.linkedin.com/posts/dietrich-gebert-b3a314a9_ever-met-a-senior-dev-whose-entire-personality-share-7471201836088164352-UVQK/ sure the dev is on x too, but his announcement was on linkedin https://github.com/DietrichGebert/ponytail

0 reply

0 recast

1 reaction

Interesting looking error message from Codex today. Wonder what could have caused it.

1 reply

0 recast

6 reactions

(1/2) 17.9% of LLM "optimized" code silently produces wrong outputs. Not crashes. Not errors. Just quietly wrong. A new paper (ComPilot, PACT 2025) ran the experiment directly: LLM proposes code transformations → compare output to original → if it matches, call it correct. 18% of the time, it matched on the test input and failed on random inputs. This is the core problem with "just ask the LLM to verify itself." It can't. What actually worked: give the LLM a real environment. It proposes a transformation, the compiler checks legality formally, the program runs, the speedup is measured. That number goes back to the LLM.

1 reply

0 recast

5 reactions

Anyone played around with this yet? https://www.cnbc.com/2026/06/26/china-zhipu-z-ai-open-source-anthropic-openai.html

3 replies

0 recast

2 reactions

who do you link to for normies about llms? your “just follow this guy and you’ll have a clue”? currently thinking Nate B might be it, like this short that gets at the stakes in local v cloud race. the right amount of YouTube dramatics with actual insight? https://youtube.com/shorts/R9qzlpFWPxw nominations please

1 reply

0 recast

1 reaction

Maybe this is what people don’t realize. Block access to frontier, and we’ll hit it with open source models. And guess what? Offensive vs defensive. The bad guys will be ready faster than you think. https://x.com/rauchg/status/2071047674187714830?s=46&t=iLSaYO2O7t6yPdmEXeqMdA

1 reply

0 recast

4 reactions

Looks like this website has been vibe clauded. I might be wrong, but once you vibe code a website, you just can’t unsee the design patterns. https://midi-visualizer.nathanielschool.com/

1 reply

0 recast

3 reactions

Turned Daniel Miessler's "prompts to run when fable comes back" into a portable agent skill. Most prompts improve one output. These improve the thing that produces all your output: your agent harness, your security posture, your strategic focus, the durability of your whole setup. 17 prompts across harness optimization, security, life and work, development, and public presence. The part most prompt lists skip, and the part I actually care about, is the running layer: when to run each one (by trigger, not on a calendar), how to run them without wrecking your config (apply reversible changes and show diffs, ask before anything irreversible), and keeping it portable across agents. Works on Claude Code, Codex, and anything that reads skills. I added a 17th prompt of my own: reconcile the configs across the different agents you run, so they stop drifting apart and contradicting each other. MIT licensed. Full credit to Miessler for the framework. github.com/kazani-351/system-upgrade

0 reply

9 recasts

13 reactions

Open. Decentralized. We need to leverage this to make AI available to all. As we’ve seen with the USG - Anthropic has Mythos with some now. OpenAI is doing the same with 5.6. USG is too powerful to control such compute frankly. Open. Decentralized. We need to make it work and win.

1 reply

0 recast

6 reactions

when i was trying to build for agentic commerce last year, i had only two ideas: 1. making reversibility/chargebacks work with stablecoins 2. "optimistic" purchasing where agents find things you would likely want/need and it shows up at your door without you even knowing i think the latter is still interesting especially as i wrap up prime day shopping

1 reply

0 recast

1 reaction

Trump style "capitalism"

0 reply

0 recast

7 reactions