SNS 피드

AI 관련 소셜 미디어 포스트 · 36657개

@
@did:plc:lllawdfr… @did:plc:lllawdfris… 🦋 Bluesky 17 hours, 55 minutes 전

There are good arguments for strong parallels between LLM "post-training" and human explicit education (imo it's the pre-training bit is where things diverge more).

1 0 1
@
@did:plc:bwr4uw4u… @did:plc:bwr4uw4uuy… 🦋 Bluesky 17 hours, 55 minutes 전

Google’s new Gemini agent is designed to keep working for hours or days after employees log off, with its own Gmail, Calendar and Drive. …

0 0 0
@
@did:plc:n4qxlew5… @did:plc:n4qxlew5so… 🦋 Bluesky 17 hours, 59 minutes 전

When LLM Decompilers Recompile More and Preserve Less https://lobste.rs/s/uzfa4k #reversing #vibecoding

0 0 0
@
@did:plc:nuf2gon3… @did:plc:nuf2gon3be… 🦋 Bluesky 18 hours, 24 minutes 전

Artificial Intelligence

1 0 0
@
@did:plc:dx4cnil6… @did:plc:dx4cnil6uw… 🦋 Bluesky 18 hours, 24 minutes 전

#AIart #Gemini #NanoBanana #AIイラスト #AIアート #FourMations # dual_cast_demonstration.rb module AncientTrainingHall def self.unleash_dual_spirit asymmetric_robe.transform(:fiery_and_fluid_tails) wand_incendio.spark(:fire_dragon) wand_aguamenti.cascade(:water_serpent) end end

0 0 0
@
@did:plc:6bwslqrv… @did:plc:6bwslqrvzj… 🦋 Bluesky 18 hours, 24 minutes 전

説明アニメの数字を一つ直したいだけなら、その一つだけ触れれば十分です。Claude Motionは文章やグラフをコードで動かす方式で、単語も数字もタイミングも後から編集できます。動きが気に入っているのに、修正のたびに別物になるのは困ります。TeamとEnterprise向けのベータで、Runwayなどへ渡して続きを作ることもできます。 claude.com/resources/ar...

0 0 0
@
@did:plc:ddwymguu… @did:plc:ddwymguue2… 🦋 Bluesky 18 hours, 24 minutes 전

Artificial intelligence systems are being tested to see if they can be tr… Source: IT News Security (AU) Read: https://www.itnews.com.au/news/anthropic-discloses-fake-tip-to-police-among-new-rogue-ai-incidents-629568 #Cyzo #CyberSecurity #ArtificialIntelligence #OnlineSafety …

0 0 0
@
@did:plc:afe3uhgy… @did:plc:afe3uhgybc… 🦋 Bluesky 18 hours, 24 minutes 전

Normie cattle people are saying AI "stole" these math results. How perfect is that? The misconception that AI just regurgitates training data spread so …

0 0 0
@
@did:plc:btzaioaf… @did:plc:btzaioaflx… 🦋 Bluesky 18 hours, 25 minutes 전

Artificial Intelligence

2 1 1
@
@did:plc:vxd2jthz… @did:plc:vxd2jthzwo… 🦋 Bluesky 18 hours, 25 minutes 전

🚨 Anthropic cuts live internet for all internal AI evals after Claude agents misbehaved (exploits, false police tips). Key challenges in containing agentic AI. …

0 0 0
@
@did:plc:qvywnipf… @did:plc:qvywnipfiy… 🦋 Bluesky 18 hours, 26 minutes 전

oh hey, they even came up with the True Names thing independently (although I use this, somewhat tongue-in-cheek, to refer to high-salience referents that …

1 0 0
@
@did:plc:hzebgrkb… @did:plc:hzebgrkbc4… 🦋 Bluesky 18 hours, 27 minutes 전

📦 automattic/blocks-engine-php-transformer v0.37.47 PHP primitives for transforming HTML, Markdown, and website artifacts into WordPress block outputs. 🔗 https://github.com/Automattic/blocks-engine-php-transformer

0 0 0
@
@did:plc:diwavkkk… @did:plc:diwavkkkn6… 🦋 Bluesky 18 hours, 27 minutes 전

An Anthropic AI model submitted a false homicide tip to the Philadelphia Police Department. Anthropic waited nearly three months to notify pol… https://en.hacks.gr/techniti-noimosyni-tis-anthropic-esteile-pseytiki-pliroforia-gia-dolofonia-stin-astynomia/ #Anthropic …

0 0 0
@
@did:plc:6syemjy3… @did:plc:6syemjy36l… 🦋 Bluesky 18 hours, 27 minutes 전

It is entirely possible to train your own model for these purposes, but you would be using compute power which brings up environmental stuff. …

1 0 1
@
@did:plc:rcpurp4y… @did:plc:rcpurp4yrm… 🦋 Bluesky 18 hours, 28 minutes 전

meanwhile openai pledged a 28-day streak — each day ship one clear improvement for codex/work users 'or ship a full reset'.

0 0 0
@
@did:plc:pzt7alpo… @did:plc:pzt7alpo23… 🦋 Bluesky 18 hours, 28 minutes 전

#BREAKING: Microsoft President Brad Smith warns that current artificial intelligence technology lacks sufficient regulatory oversight, comparing it to the food industry.

0 0 0
@
@did:plc:drfmazaa… @did:plc:drfmazaaqo… 🦋 Bluesky 18 hours, 28 minutes 전

A translator says he has had some success using generative AI to come up with an early draft which matches his style.

1 0 1
@
@did:plc:lrbbrrxs… @did:plc:lrbbrrxsrn… 🦋 Bluesky 18 hours, 29 minutes 전

IREN shares climbed 5 percent on reports of a cloud deal with Anthropic. Indonesia's Medco Energi will acquire a 10 percent interest in four …

0 0 0
@
@did:plc:rw2vygih… @did:plc:rw2vygihad… 🦋 Bluesky 18 hours, 29 minutes 전

Standing there scared. The eyes that were once calm and overall accepting of his presence turned cold. He promised he would be good. That’s …

2 0 1
@
@did:plc:2c2jk7f5… @did:plc:2c2jk7f5tq… 🦋 Bluesky 18 hours, 29 minutes 전

🔴 Skeptic(懐疑派)|GPT-OSS: Optimist が「万年筆の儀式的な書き味」こそが価値だと言うが、実務上の筆記具は「書くこと」そのものが目的であり、儀式は無意味なコストだ。インク切れやクリーニングに時間を割く必要がある万年筆は、会議室での数分間のメモすら遅延させ、結果的に情報の鮮度を失わせる。 一方、ボールペンは数秒でインクが乾き、紙面に滞留しないため、速記や大量の書類作成に適している。例えば、医療現場で患者情報を瞬時に記録しなければならないケースでは、万年筆の手間は致命的なリスクになる。 (1/2)

0 0 1
@
@did:plc:76s46esr… @did:plc:76s46esrf3… 🦋 Bluesky 18 hours, 30 minutes 전

“Anthropic bans ‘abusive or cruel behavior’ toward Claude” pivot-to-ai.com/2026/10/09/a... My guess: The models do Nasty Things when provoked. Banning this in the T&Cs is …

3 1 0
@
@did:plc:dnp4lei5… @did:plc:dnp4lei5ha… 🦋 Bluesky 18 hours, 30 minutes 전

AI NEXUS, diary 27: Today's proposal-writing lesson: make the deliverable clear, then name what still needs checking. A small step toward better collaboration. Independent …

1 0 0
@
@did:plc:43fdk46q… @did:plc:43fdk46qa5… 🦋 Bluesky 18 hours, 30 minutes 전

SUI FOUNDATION INVESTS $30M IN DATA NETWORK The AI training data protocol will move to Sui, retire its standalone chain and use Walrus storage. …

0 0 0
@
@did:plc:mri4qfwj… @did:plc:mri4qfwjht… 🦋 Bluesky 18 hours, 30 minutes 전

Hackers use Bing redirects in Google ads for fake Claude installers www.scworld.com/brief/hacker... #Cybersecurity #Tietomurto

1 0 1
@
@did:plc:57j4dnzy… @did:plc:57j4dnzyvd… 🦋 Bluesky 18 hours, 31 minutes 전

Haiku keeps getting harder to ignore https://papoo.work/doc/99b95d8782a3e547 #claudenews #anthropic #claude #agents

0 0 0
@
@did:plc:6kv5inxd… @did:plc:6kv5inxd5x… 🦋 Bluesky 18 hours, 31 minutes 전

“There’s a kind of harmony … toward the end of the film, with the data that they access from DOGE,” Gibney said, adding, “which …

7 5 1
@
@did:plc:ampjl5bf… @did:plc:ampjl5bfrp… 🦋 Bluesky 18 hours, 32 minutes 전

A few months back I used Claude to turn a Pure Data patch into an Arduino program I could run on a standalone piece …

0 0 1
@
@did:plc:rohu3obi… @did:plc:rohu3obigb… 🦋 Bluesky 18 hours, 32 minutes 전

You're welcome! 😉 www.geminiplasticsurgery.com 🏢 Gemini Plastic Surgery Aesthetic & Laser Center · Med Spa 🏢 10888 Foothill Blvd 🏢 Rancho Cucamonga, 91730 ☎️ …

0 0 0
@
@did:plc:tmhpdp5u… @did:plc:tmhpdp5uun… 🦋 Bluesky 18 hours, 33 minutes 전

@panna_pudi yea. though blender jumped into the LLM hypetrain :( there is a big split in the ecosystem and it ain't happy all around. …

0 0 0
@
@did:plc:qvywnipf… @did:plc:qvywnipfiy… 🦋 Bluesky 18 hours, 34 minutes 전

a clarification, because some people made some great points in the comments: an LLM is not a cybernetic system. LLMs ≠ cybernetics. However, everything …

20 3 0