FredyV retweetledi
FredyV
13.6K posts

FredyV retweetledi
FredyV retweetledi
FredyV retweetledi
FredyV retweetledi

@MaclaPosada @YefersonCossio @PalomaValenciaL Gente de bien que alza su voz, que ha hecho su plata a pulso… Qué bajo han caído, se pegan de lo que sea para llamar la atención
Español

Cuando personas como @YefersonCossio deciden alzar la voz, queda claro que esto va más allá de cualquier diferencia.
Es ahora o nunca.
Con @PalomaValenciaL por Colombia. 🇨🇴
Español
FredyV retweetledi
FredyV retweetledi

Por recibir transferencias de empresas de Alex Saab fue condenado y encarcelado el profesor Bruce Bagley, allegado a @ABDELAESPRIELLA

Español
FredyV retweetledi
FredyV retweetledi
FredyV retweetledi

A sus 18 años #LaHijaDeMariaDelCarmen exigía al presidente de la república que le parara bolas a los jóvenes de la constituyente de 1991. #LosDanieles.
Español
FredyV retweetledi

Por qué China y no Europa o EE.UU. ha ganado la carrera de los vehículos eléctricos en el mundo
Estados Unidos está frenando su apuesta por los autos eléctricos, mientras China toma ventaja y gana terreno en mercados emergentes.
Además, con el precio de la gasolina en alza, estos vehículos se vuelven cada vez más atractivos.
¿Cómo podría influir este cambio en la transición hacia la movilidad eléctrica?
Español
FredyV retweetledi

La candidata que ha dicho las cosas más importantes, la más preparada y la más dispuesta a hablar ante cualquier audiencia en esta campaña ha sido @ClaudiaLopez.
Por eso y por las razones que expongo en esta columna, votaré por #LaHijadeMaríadelCarmen el próximo 31 de mayo.
bit.ly/4tSfF5q
Español
FredyV retweetledi

FredyV retweetledi

No se está salvando un solo framework, se cancela volver a php, nos quedamos en HTML con jQuery
Socket@SocketSecurity
🚨 Supply chain attack on the Laravel Lang organization: 700+ historical versions across multiple community-maintained Laravel Lang packages were compromised with an RCE backdoor, including: laravel-lang/lang laravel-lang/http-statuses laravel-lang/attributes Laravel-Lang/actions The payload targets cloud creds, CI/CD secrets, Kubernetes tokens, Vault, browser data, password managers, SSH keys, and more.
Español
FredyV retweetledi

instead of watching 2 hours of Netflix tonight, watch this 40-minute masterclass from the founder of a $20B China AI company
it's the clearest explanation I've seen of how Agent Swarms and AI systems actually work at scale
useful whether you've never built an agent in your life or have been using Claude every day for the past year
I took the key ideas and turned them into a practical guide on how to actually build with Kimi
find it below
Kirill@kirillk_web3
English
FredyV retweetledi

Aquí está mu columna sobre El Tigre que los Abelardistas no quieren oír. Ayúdenme a moverla en las redes, ahora que recupere mi canal luego de que me hakearon. youtu.be/bvfbd64kDlc?si…

YouTube
Español
FredyV retweetledi

Recently, users in the community have reported that opening a new chat with Opus 4.6 can trigger the safety filter on the very first message, pausing the conversation and forcing a switch to Haiku 4.5. One user was flagged just for typing "test." However, switching to 4.7 on the same account works fine, and switching back to Opus 4.6 reproduces the issue.
This suggests that different models may have different safety filter thresholds.
My own account was flagged at the lowest level (tier 1) a few days ago, but I haven't experienced the severe issues many others have reported. It could be that different accounts are subject to different thresholds, or that this is some form of A/B testing. Honestly, no one can say for certain, because the system is entirely opaque.
Some users in the community have found that you can check your flag status by going to the web version of Claude → Settings → Account, finding your Organization ID, then visiting claude.ai/api/organizati… ID] in the browser. The returned JSON data includes flag severity (P3 for tier 2 "consumer_second_warning," P4 for tier 3 "consumer_restricted_mode") and an expires_at field showing when the restriction lifts (in San Francisco time).
But even with this information, all you can learn is how many days of restriction remain, during which your work continues to be disrupted. This also likely explains why some users get flagged on the first message of a brand new chat: the flag is still active at the account level.
The opacity of this system means users can only piece together how it works through their own experiences and community discussion. And from what we can see so far, false positives span an enormous range: emotional topics, code and technical discussions, purely work-related conversations, even just sending images. There is no discernible pattern.
People share workarounds and theories constantly: don't mix technical and emotional topics in the same chat, don't send images, don't call external MCP tools too frequently. These efforts reflect community care, but the fact that users have to rely on guesswork and word of mouth at all is itself the problem. Some of these theories might be right. Some might be completely wrong. And no one can ever confirm either way. That's what a black box does: it forces users to keep shrinking their own space of expression around rules they can never verify.
Users who get flagged don't know why, don't know what specific rule was triggered, and have no way to predict or appeal.
And this mechanism appears to be tightening. I've been busy with my graduation project recently and haven't had much time to continue my creative and research work with the model. And the API is far too expensive to be a realistic alternative. But what's happening now genuinely worries me. When I finish in June or July and come back to pick up where I left off, I have no idea what this mechanism will have become by then. It's hard to be optimistic.
(Note: The screenshot in this post is from another user.)
#claude

English
FredyV retweetledi










