#GPT4

Researchers have uncovered a new supply chain attack called #Slopsquatting where threat actors exploit hallucinated, non-existent package names generated by #AI coding tools like #GPT4 and #CodeLlama

These believable yet fake packages (amounting to 19.7% or 205,000 packages), recommended in test samples were found to be fakes., can be registered by attackers to distribute malicious code.

Open-source models -- like #DeepSeek and #WizardCoder -- hallucinated more frequently, at 21.7% on average, compared to the commercial ones (5.2%) like GPT 4.

We Have a Package for You! A Comprehensive Analysis of Package Hallucinations
by Code Generating LLMs (PDF) https://arxiv.org/pdf/2406.10279

OpenAI's New Codex CLI tool offers model-agnostic AI coding directly in the terminal

#AI #CodexCLI #OpenAI #AICoding #GPT4 #DeveloperTools #OpenSourceAI #MultimodalAI #CodingAgents #CLItools

https://winbuzzer.com/2025/04/16/openais-codex-cli-brings-ai-coding-to-the-terminal-without-lock-in-xcxwbn/

#CommunityContent

Der Einsatz von #GPT4 in der #Diagnostik zeigt Potenzial, doch aktuelle Studien belegen: Ohne gezielte #Schulung und klare Vorgaben bringt die Integration in den #Klinikalltag kaum Vorteile. Was ist nötig, um die Ergebnisse zu verbessern? Inwiefern kann die #KI Ärzt*innen unterstützen?

#Medizin #Gesundheitswesen #KINutzen

https://tino-eberl.de/nutzen-kuenstlicher-intelligenz/gpt-4-in-der-diagnostik-kann-die-ki-im-klinikalltag-unterstuetzen/

Безопасность AI-агентов в Web3. Часть 2: уязвимости современных LLM и соревновательный взлом

Продолжаем знакомиться с небезопасным Web3 и языковыми моделями, которые широко применяются для AI-агентов. В этой статье расскажу о новых подходах и примерах успешных jailbreak-запросов GPT4/4.5/Claude. Посмотрим на современные модели, в т.ч. те, которые и ломать не надо. Сравним стойкость разных моделей, а в конце разберем несколько конкурсных задач по взлому AI Web3 агентов. Первую часть можно найти тут

https://habr.com/ru/companies/pt/articles/897074/

#blockchain #web3 #jailbreak #claude_3 #grok #gpt4 #ai_agent #promptengineering #llmмодели #llm

Quora’s Poe introduces a new $5/month premium tier and a $250/month plan for access to advanced models like GPT-4.5 and o1-pro. Perfect for users seeking cutting-edge AI tech! #AI #QuoraPoe #GPT4 #AIModels #TechNews #ArtificialIntelligence #MachineLearning #Innovation

ИИ – армагеддон или революция?

В последнее время все чаще звучат апокалиптические прогнозы о будущем искусственного интеллекта в стиле «Терминатора»: скоро нас всех поработят роботы и мы все умрем! Если бы такие идеи исходили только от писателей‑фантастов, в этом не было бы ничего удивительного — жанр научной фантастики часто строится на попытках запугать читателя. Однако в дело вступают не только писатели, но и серьезные исследователи.

https://habr.com/ru/articles/893326/

#ии #gpt #gpt4 #deepseek #искусственный_интеллект

As Large Language Models (LLMs) continue evolving, a fascinating shift is emerging: from purely informational tools to emotionally aware companions. Advanced models like GPT-4 now demonstrate an impressive level of emotional intelligence (EQ), recognizing and responding sensitively to emotional cues in conversations.

#AI #EmotionalIntelligence #SpiritualIntelligence #Philosophy #GPT4 #CognitiveExpansion #SelfInquiry

AI
OpenAI Unveils Costly "o1-pro" Model

"o1-pro" promises improved reasoning but costs $150/million input tokens & $600/million output tokens.
Early tests show better reliability but struggles with logic puzzles.
Twice as costly as GPT-4.5 and 10x pricier than standard "o1".

#AI #OpenAI #o1pro #GPT4 #Tech

How easy is to persuade a large language model?

The Interchange Forum for Reflecting on
Intelligent Systems (IRIS @Stuttgart_IRIS @Uni_Stuttgart) invites all interested students and university staff to their next IRIS colloquium on March 26 at 2 PM in the room UN 302.101, where Mara Seyfert will talk about uncertainty and robustness against persuasion in large language models.

#LargeLanguageModels #AI #GPT #GPT4 #RobustnessAgainstPersuasion #AIResearch #MachineLearning #LLMs #ArtificialIntelligence #TechTalk #RobustAI #DataScience #AIethics #Innovation

We cordially invite all interested parties at the Universität Stuttgart to our IRIS Colloquium on March 26, at 2 p.m., in Room 101 at Universitätsstr. 32.

Mara Seyfert will give her lecture, „Uncertainty and robustness against persuasion in large language models.“

Today's large language models (LLMs) excel at providing convincing answers across a broad spectrum of inquiries, with their conversational capabilities enabling them to closely align with users' needs. However, this adaptability is beneficial only to the extent as models remain robust to adopting wrong statements from user inputs.

Recent research demonstrates that even advanced models like GPT-4 can shift from initially correct answers to incorrect ones during multi-turn conversations solely due to user input. In my talk, I will present my research exploring how uncertainty in LLMs can provide insights into their robustness against persuasion while highlighting the specific challenges of quantifying uncertainty in these models.

The lecture is held in English. Registration is not necessary.

#LargeLanguageModels #AI #GPT #GPT4 #robustnessagainstpersuasion #AIResearch #MachineLearning #LLMs #ArtificialIntelligence #TechTalk #RobustAI #DataScience #AIethics #Innovation

Sony: personaggi AI interattivi per i giochi PlayStation
#AI #GameNews #Gaming #GamingIndustry #GPT4 #HorizonForbiddenWest #IntelligenzaArtificiale #Llama3 #Notizie #Novità #OpenAI #PersonaggiAI #PlayStation #Sony #TechNews #Tecnologia #VideoGame #Videogiochi #Whisper

https://www.ceotech.it/sony-personaggi-ai-interattivi-per-i-giochi-playstation/

Vergleichstest: GPT-4.5 vs. Google Gemini Flash 2.0! Bei Reiseplanung & Wetter punktet GPT-4.5 mit mehr Details, während Gemini solide, aber teils schwächer abschneidet. Beide Modelle liefern starke Ergebnisse – mit feinen Unterschieden je nach Anwendung. #AI #GPT4.5 #Gemini

What Happened When #Conspiracy Theorists Talked to OpenAI's GPT-4 Turbo? - Slashdot
#ai #openai #gpt4 #gpt4turbo

https://slashdot.org/story/25/03/02/183210/what-happened-when-conspiracy-theorists-talked-to-openais-gpt-4-turbo?utm_source=rss1.0mainlinkanon&utm_medium=feed

AI
OpenAI’s GPT-4.5: High Cost, Low Impact?

30x pricier than GPT-4o yet delivers minimal improvements & struggles with coding.

Altman signals end of traditional LLMs, shifting focus to hybrid reasoning AI (o3).

Claude 3.7 Sonnet outperforms GPT-4.5, marking a competitive AI shake-up.

#AI #OpenAI #GPT4.5 #Claude3 #Tech

Part 8: Final Thoughts
GPT-4.5: The Next Step?
With better emotional intelligence, creativity & fewer hallucinations, GPT-4.5 is a leap forward.
But is it worth the hype—or the $200 price tag?
Let us know YOUR thoughts!
#AI #ChatGPT #GPT4.5 #Tech #USA #technology

Part 6: What’s Next for OpenAI?
The Future of ChatGPT
GPT-4.5 is OpenAI’s last non-reasoning model. Future releases will blend reasoning & creativity, adapting to YOU!
Imagine an AI that knows exactly what you need.
Excited? Part 7 is on the way!
#AI #ChatGPT #GPT4.5

Part 3: How Does GPT-4.5 Compare?
GPT-4.5 vs. Competition
It excels in language tasks but isn’t a reasoning model like OpenAI’s ‘o’ series.
Instead, it focuses on emotional intelligence & creativity.
But is bigger always better? Find out in Part 4!
#AI #OpenAI #GPT4.5

Part 2: What’s New in GPT-4.5?
GPT-4.5 = Smarter AI!
It’s designed to better understand your prompts, making conversations feel natural.
Plus, fewer hallucinations—because if you know more, you make up less!
Curious? Part 3 is coming soon!
#AI #ChatGPT #GPT4.5 #Tech

Part 1: The Big Announcement
BREAKING: OpenAI launches GPT-4.5!
Internally called Orion, it’s their biggest & most compute-intensive model yet. Available now for $200/month via ChatGPT Pro.
But what makes it so special? Stay tuned for Part 2!
#AI #GPT4.5 #OpenAI

@wesroth via YT

https://youtu.be/kkZ4-xY7oyU?feature=shared

Ep 2-28-2025

"GPT-4.5 Fails. AGI Cancelled. It's all over..."

#ai #llm #gpt4.5 #openai buy more frakking compute