hhardy01 (@hhardy01@mastodon.social)
Link to new Sora-made teaser trailer for the upcoming episode of "The Starry Messenger" titled, "The Dog Who Found the Broken God".
This is an upcoming episode of my stories, Traveller-compatible RPG, and persistent world of "The Starry Messenger."
Made using ChatGPT 5.1, Sora 2, Audacity, Adobe Express and Google video.
TBH this isn't near broadcast quality but its coming along nicely.
What do you think?
https://drive.google.com/file/d/1v0U7nBzZgIXHPx8U_vA7q5XEg49J0IMs/view?usp=vids_web
#art #aiart #video #Sora #ChatGPT #mastoart #sciencefiction
Show Original Post
hhardy01 (@hhardy01@mastodon.social)
Link to new Sora-made teaser trailer for the upcoming episode of "The Starry Messenger" titled, "The Dog Who Found the Broken God".
This is an upcoming episode of my stories, Traveller-compatible RPG, and persistent world of "The Starry Messenger."
Made using ChatGPT 5.1, Sora 2, Audacity, and Adobe Express.
TBH this isn't near broadcast quality but its coming along nicely.
What do you think?
https://drive.google.com/file/d/11yZc_Vceqx2LdaYtVm7q7RiXNkBMUCRe/view?usp=drive_link
#art #aiart #video #Sora #ChatGPT #mastoart #sciencefiction
Show Original Post
poorpossum (@poorpossum@kolektiva.social)
Look, as a child of the 90s I'm treating AI like a nicotine product: Just fucking don't.
#AI #GenAI #ChatGPT #BetterOffline #Addiction #AIPsychosis
Show Original Post
ycombinator (@ycombinator@rss-mstdn.studiofreesia.com)
Prompt Injection via Poetry
https://www.wired.com/story/poems-can-trick-ai-into-helping-you-make-a-nuclear-weapon/
#ycombinator #artificial_intelligence #machine_learning #algorithms #nuclear_war #chatgpt #openai #meta #anthropic
Show Original Post
aistorynews (@aistorynews@techhub.social)
AI browser startups promised effortless agents, yet tests show accuracy and UX gaps as Edge Copilot and Chrome Gemini set tougher standards.
#AIHardware #Automation #ChatGPT
Show Original Post
Hackread (@Hackread@mstdn.social)
Command copied, control lost: Fake #ChatGPT Atlas Browser used in a #ClickFix attack tricking users into installing password‑stealing malware via a fake installer.
Read: https://hackread.com/fake-chatgpt-atlas-clickfix-steal-passwords/
#CyberSecurity #Phishing #Malware #ChatGPTAtlas #Infosec
Show Original Post
remixtures (@remixtures@tldr.nettime.org)
"OpenAI chief executive Sam Altman has declared a “code red” over the need to improve ChatGPT, as rivals Google and Anthropic narrow its early lead in the race to develop artificial intelligence.
According to an internal memo sent on Monday, Altman said the code red “surge” would refocus OpenAI’s work on ChatGPT, such as improving the speed, reliability and personalisation of the market-leading chatbot.
He told employees the $500bn start-up was planning to delay other products as “we are at a critical time for ChatGPT”.
That means the company will push back plans to develop advertising products, AI agents intended to automate shopping and health-related tasks, and Pulse, which produces personalised morning updates for users.
The Information first reported details of the memo.
The strategic overhaul comes as OpenAI grapples with growing competition, rising data centre costs, the technical challenges of remaining at the frontier of AI and the constant battle to retain key talent.
Last month, Google released its latest large language model, Gemini 3, which is considered to have leapfrogged OpenAI’s GPT-5 on industry benchmark tests. Anthropic’s latest model, Opus 4.5, also outperformed GPT-5 in key benchmarks."
https://www.ft.com/content/7a42396f-487a-47b0-8121-8d8f2112fa53
#AI #GenerativeAI #OpenAI #AIBubble #ChatGPT #LLMs
Show Original Post
r (@r@fed.brid.gy)
ChatGPT Told a Violent Stalker to Embrace the 'Haters,' Indictment Says
Show Original Post
Bluedonkey (@Bluedonkey@mastodon.social)
Thoughts for the morning... are our current #AI models more or less efficient than the human brain?
To be clear, I don't just mean running queries, which is clearly more efficient in both speed and energy consumption, but including the energy to train them, acquire the training materials and develop the models. Seems that we might not have reached the point where they are as efficient when the whole lifecycle is included.
#ML #chatgpt #openai #googleAi #anthropic
Show Original Post
display (@display@connexia.hibiol.eu)
O CNRS (Centro Nacional de Pesquisa Científica), maior órgão público de pesquisa científica da França e uma das mais importantes instituições de pesquisa do mundo, anunciou o lançamento do Emmy, um agente conversacional em inteligência artificial generativa.
Eles fizeram um acordo com a Mistral, concorrente francês da OpenAI, cujos valores não foram divulgados, para que, entre outros, seus agentes não utilizem outros chatbots, o que compromete a segurança dos dados do CNRS.
Acessível através de um portal do centro de pesquisa a partir de 16 de dezembro de 2025, este chatbot utilizará os modelos da Mistral AI, “graças a um acordo firmado” com a empresa, explica a DSI do CNRS.Em sua primeira versão acessível aos agentes, destacam-se as seguintes funcionalidades:
- tradução de textos em todas as línguas;
- síntese de documentos;
- ajuda na reformulação;
- ajuda na reflexão;
- pesquisa na web;Mistral AI
- reconhecimento de textos e imagens;
- modo “raciocínio”: a IA trata a questão do usuário passo a passo para dar uma resposta mais pertinente e completa;
- coleções de documentos: permite adicionar documentos externos para contextualizar a reflexão (por exemplo, um pesquisador pode baixar suas publicações para permitir que a IA contextualize e enriqueça suas respostas).
O nome feminino (como Tay, Siri ou Alexa) usado para o chatbot é apresentado como uma referência à matemática alemã Emmy Noether.
Em sua mensagem, a DSI explica que um “módulo de sensibilização para a inteligência artificial, bem como recomendações para o uso da IA generativa no CNRS” serão propostos aos agentes “até 16 de dezembro para garantir o uso da Emmy em um ambiente seguro e adequado”.
Acrescenta que será necessário apresentar um certificado de conclusão deste módulo para aceder à Emmy e que «tendo em conta os riscos comprovados que outras ferramentas de IA na nuvem pública representam para a confidencialidade dos dados e a proteção do patrimônio informativo da instituição, qualquer outra ferramenta de agente conversacional em IA generativa será doravante proibida no CNRS».
A direção do centro de pesquisa, no entanto, não fornece informações a esses agentes sobre o orçamento dedicado ao acordo firmado com a Mistral AI.
next.ink/brief_article/le-cnrs…
#mistral #mistralAI #france #chatbot #chatgpt
Show Original Post
13 (@13@2137.social)
In Which I Vibe-Code a Personal Library System https://hackaday.com/2025/12/03/in-which-i-vibe-code-a-personal-library-system/
#ArtificialIntelligence #Featured #Interest #OriginalArt #Barcode #Barcodescanner #ChatGPT #Flask #Library #Librarysystem #SQLAlchemy #Vibecoding
Show Original Post
carstenfranke (@carstenfranke@mastodon.social)
The reason I post these #aifailures is that they are easy to comprehend. It is obvious, to most people, that they are nonsense. The #LLM s create the same nonsensical answers for more complex questions, but since people do not know the real answer to these questions, they do not realize that what the #ai is giving them is wrong.
#chatgpt #gemini
Show Original Post