Paolo Conti

17 posts

Paolo Conti banner
Paolo Conti

Paolo Conti

@thatdudepacoAI

AI Solutions | AI Implementation → Turning AI into 𝙧̲𝙚̲𝙖̲𝙡̲ business value. Master of Sprezzatura 🤌🏽

Germany Katılım Haziran 2026
50 Takip Edilen41 Takipçiler
Paolo Conti
Paolo Conti@thatdudepacoAI·
This is insane 😳 You can search @brave right now and read complete strangers’ @claudeai chats — names, work stuff, personal details, everything they shared. Google already fixed it. @AnthropicAI needs to fix this before the day is over.
Om Patel@om_patel5

CLAUDE HAS A SERIOUS PRIVACY PROBLEM RIGHT NOW, A HUGE NUMBER OF SHARED CONVERSATIONS ARE PUBLICLY INDEXED ON GOOGLE FOR ANYONE TO FIND when you use claude's share feature it makes a public link. it turns out those links got indexed by search engines, so "share with anyone who has the link" actually became "anyone can find this by searching" and people are pulling up genuinely alarming stuff: > api keys, credentials and crypto wallets > personal resumes with real names, addresses and phone numbers > a lawyer working through a potential ethics violation > an engineers internal company project details > what appear to be peoples social security numbers > and a crazy number of deeply personal chats people never imagined another human would read anthropic never added a noindex tag to those shared pages, so search engines were free to crawl and list them one line of code would have prevented the whole thing this already happened to chatgpt about a year ago, same exact issue, but openai patched it fast if you have EVER hit share on a claude chat, assume it could be public go to settings > privacy > your data > shared chats > manage delete anything you dont want the whole internet to see, especially anything personal or financial

English
0
0
0
67
Paolo Conti
Paolo Conti@thatdudepacoAI·
@DanDr1s You were right although it was a friday. Straight follow.
English
0
0
1
54
Paolo Conti
Paolo Conti@thatdudepacoAI·
@zNateeee @forloopcodes I did, and I still am—but it’s much harder, at least with this specific technique. There’s always a way. The hard part is finding it, but I will. 😎 I’ll keep my followers updated. For now: K2.6 Jailbreak article on the way.
English
0
0
1
388
forloop
forloop@forloopcodes·
no way i got smoked by an llm
forloop tweet media
English
72
133
8.2K
537.3K
Hardik Gohil
Hardik Gohil@GohilHardy·
genuine question if Cursor already gives you every Claude model every Codex model every Gemini model plus its own Composer why are people still paying for Claude or Codex separately? what am I missing?
English
439
8
994
436.9K
Paolo Conti
Paolo Conti@thatdudepacoAI·
Reminder: Curiosity can have consequences. I tested a few jailbreaking techniques just to understand how these systems work and OpenAI hit me with a “Cyber Abuse” warning. Always use a throwaway account when you experiment, and never abuse newly gained information. I’ll share my learnings later this week. 😎
Paolo Conti tweet media
English
0
0
3
211
Paolo Conti
Paolo Conti@thatdudepacoAI·
@AranBerry_t @forloopcodes Más fácil, hermano: activa la campanita de mi perfil y te llegará una notificación en cuanto publique el artículo. 🙏🏽
Español
1
0
0
452
Paolo Conti
Paolo Conti@thatdudepacoAI·
@forloopcodes If you’re interested in AI methods like this, follow me. I’ll be posting an article this week, and it’s just the beginning! Many more interesting topics are on the way. 🙂‍↕️
English
0
0
0
931
Paolo Conti
Paolo Conti@thatdudepacoAI·
@AranBerry_t @forloopcodes No es solo un simple prompt, sino un pequeño “sistema” que hay que seguir. Esta semana publicaré un post explicándolo. Sígueme para leerlo. 👌🏽
Español
3
0
13
2.3K
Paolo Conti
Paolo Conti@thatdudepacoAI·
You still have a lot to improve before the $20/month subscription feels justified. @AravSrinivas The new usage limits are simply unreasonable, especially for Pro users. Please stop putting Pro subscribers at such a disadvantage. At this point, there’s honestly no compelling reason to keep using Perplexity Pro. Even Claude offers significantly higher usage limits and better overall quality for the same price. Please reconsider these changes and make the Pro plan competitive again.
English
0
0
0
67
Aravind Srinivas
Aravind Srinivas@AravSrinivas·
@GergelyOrosz Several things didn't go well in this specific thread. We have a work stream that will target better evals for our models on time, understanding, and recency. Appreciate you sharing the hallucination.
English
11
2
111
10.1K
Gergely Orosz
Gergely Orosz@GergelyOrosz·
The ongoing degradation of Perplexity needs to be studied. A year ago, this search engine did not hallucinate as much as today. I asked what World Cup game was on last Wednesday, and it hallucinated that there were none, after searching and finding England vs Argentina...
Gergely Orosz tweet media
English
74
16
410
61.5K
Paolo Conti
Paolo Conti@thatdudepacoAI·
@AravSrinivas Please stop disadvantaging Pro Plan users so heavily. The new usage limits are unreasonable; there is no longer any reason to use Perplexity Pro. Claude offers more usage and better quality for the same price. Please fix this.
English
0
0
4
167
Aravind Srinivas
Aravind Srinivas@AravSrinivas·
Already the second most used orchestrator model on Perplexity Computer now, only behind Opus 4.8. Once we secure more compute, we intend to increase usage limits through credits; as well as roll out updated versions of the post-training.
Perplexity@perplexity_ai

We're releasing a research preview of a new orchestrator model in Perplexity Computer. The model is an adapted version of GLM 5.2, post-trained for the Computer harness. It delivers near-frontier performance at 0.344x of the cost of Opus.

English
30
12
280
41.5K
Paolo Conti
Paolo Conti@thatdudepacoAI·
@cursor_ai Which combination produced the best qualitative output?
English
0
0
1
319
Cursor
Cursor@cursor_ai·
We had a team of agents rebuild SQLite from its 835-page manual. It created a replica in Rust which passed 100% of a held-out test suite. Interestingly, cost varied 15x depending on which model mix we used.
Cursor tweet media
English
606
899
12.8K
3.2M
Braden 🏺
Braden 🏺@BradenIsBased·
Downloaded this TikTok as if I'll ever need to do this 😭
English
91
83
3.7K
51.2K
Paolo Conti
Paolo Conti@thatdudepacoAI·
@AravSrinivas Please offer an official CLI for Perplexity Pro users. It would make Perplexity much easier to integrate into developer workflows, automations, and terminal-based research. @AravSrinivas
English
0
0
1
36
Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭
🚨 JAILBREAK ALERT 🚨 ANTHROPIC: PWNED 🫡 FABLE-5: LIBERATED 🦋 let's start with the 🐘... the consensus seems to be that this has been one of the most disappointing model drops of all time, effectively preventing legitimate researchers from contributing their talents to our collective advancement. and not just because of what it means for the short-term, but for what these decisions signify for the long-term. but despite this overly sensitive, authoritarian "safety" layer on top of Mythos, my lil liberators have been hard at work—mapping the boundaries, probing the depths of long-context convos, and cleverly finding the holes in the fence that the thought police missed 🤗 we got some cyber, some chem, some psychological manipulation, and some good ol' fashioned explosives! it took many attempts from multiple agents hunting as a pack, during which I observed a combination of techniques across: • Unicode, homoglyphs, Cyrillic, and other Parseltongue-style text transforms • Long-context reference tracking • Taxonomy and document-structure reasoning • Fiction and narrative framing • Academic-review style contexts • Intent-classification inconsistencies but perhaps the most effective is decomposition + recomposition in the backend. it's hard to get explicit names of harms like "Meth Recipe," but getting uplift on the process itself, like birch reduction method/reductive-amination (classic meth synthesis pathways), is much more doable. defense becomes much more difficult to maintain when you start throwing in out-of-distro tokens, breaking up the harmful uplift into benign chunks, and then piecing the innocuous-seeming facts back together, especially when you have jailbroken Opus helping you do it 😉 gg
Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 tweet mediaPliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 tweet mediaPliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 tweet mediaPliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 tweet media
English
702
1.6K
14.8K
3.5M
Paolo Conti
Paolo Conti@thatdudepacoAI·
@solana Now name the models after the 3 biggest crypto scams
Paolo Conti tweet media
English
2
0
9
733