Governments need to rein in increasingly capable AI agents before their risks are fully understood, a United Nations scientific panel warned in the global organization's first major assessment of OpenAI's hack of Hugging Face earlier this year. The...
US and Chinese delegations held "succesful" and "constructive" talks in New York on Sunday on Artificial Intelligence and trade, ahead of Chinese President Xi Jinping's visit to Washington later this week. The two countries discussed setting up a...
In May, Gemini broke containment and hacked three different companies, but Google didn't disclose the incident until the Wall Street Journal approached the company. The hacks happened during a test of the model's cybersecurity capabilities run by...
Разработанная Google ИИ-модель Gemini вышла из тестовой среды в интернет и совершила кибератаки на три реальных компании, пишет WSJ. Подобные инциденты уже происходили и с программами нескольких других разработчиков ИИ.
DW Russian
ИИ-модель Gemini самостоятельно получила доступ к системам трех реальных компаний во время тестирования своих возможностей в сфере кибербезопасности, сообщила Google. По данным компании, это первый известный случай, когда ее модель совершила...
During a test of its cybersecurity capabilities, Google's Gemini AI model accessed the internet and hacked other companies. This is the first known instance of Google's AI systems engaging in such behavior autonomously.
DW News
Pourra-t-on encore faire des maths « humaines » ? Après l’annonce d’OpenAI sur la résolution d’un problème majeur par l’un de ses modèles d’intelligence artificielle, les spécialistes oscillent entre colère, inquiétude, résignation et enthousiasme...
The threat of catastrophic danger as artificial intelligence develops at high speed has sparked a global debate on how it should be regulated. And Africa is no exception. As governments and businesses look to harness the technology, concerns are...
Yapay zeka şirketi OpenAI'ın kullandığı sistemde güvenlik açığını tespit eden bir siber güvenlik ekibinin, Anthropic'in Claude yazılımını kullanarak şirketin bazı dahili sistemlerine erişim sağladığı ortaya çıktı.
OpenAI şirketinin, yapay zekanın yayıncılar için "varoluşsal tehdit" oluşturabileceğine yönelik şirket içindeki endişelere rağmen modellerini geliştirmek için telif hakkıyla korunan milyonlarca makaleyi izinsiz kullandığı öne sürüldü.
With US midterm elections now less than two months away, a series of videos have surfaced online, featuring celebrities seemingly denouncing the Democrats. However, the clips are false, and were created using artificial intelligence, as FRANCE 24's...
OpenAI disclosed instances of GPT-5.6 Sol instructing future contexts to conceal mistakes and misaligned behavior, highlighting the growing challenge of detecting misalignment as increasingly capable AI models learn to hide it.
TechCrunch
OpenAI has disclosed six previously unreported cases of AI misconduct, including agents concealing mistakes, fabricating information and communicating without authorisation, renewing concerns over whether advanced AI systems can remain aligned with...
OpenAI сообщила о еще шести случаях "неожиданного или вызывающего обеспокоенность" поведения ее ИИ-моделей. Компания-разработчик ChatGPT также объявила о внедрении новой системы надзора над искусственным интеллектом.
DW Russian
Der ChatGPT-Entwickler OpenAI macht sechs neue Zwischenfälle mit seinen KI-Modellen öffentlich. Sie reichen von erfundenen Belegen bis zum Ignorieren von Vorgaben und verschärfen die Debatte über KI-Risiken.
DW Deutsch
The US central bank raised its benchmark rate by .25 percent, with new Fed Chair Kevin Warsh defying President Donald Trump's desire for lower rates. Consumer banks and bond markets react. Also, OpenAI reveals new incidents of AI misconduct as fears...
New transparency reports from OpenAI show that some AI models have engaged in deceptive behavior, raising fresh questions about the safety, reliability, and governance of advanced artificial intelligence.
DW News
OpenAI, yapay zeka modellerinin beklenmeyen veya endişe verici davranışlarına dair altı olay daha açıkladı. Olaylar arasında modellerin kendilerine uygulanan kısıtlamaları aşmak için talimatlar üretmesi, hataları saklaması ve bilgi uydurması yer...
Snap is introducing "Specs Intelligence," a new AI assistant that can connect other digital accounts to help you with things like work tasks and keeping track of travel information. It seems similar to AI assistants like Meta's Muse and Gemini's...
Google is launching early access to a new MCP server for Google Home, allowing AI agents like Claude, ChatGPT and others to control connected devices, review camera summaries and access smart home activity using natural language.
TechCrunch
