EVENT DOSSIER
Three leading AI companies jointly issued a statement on August 28, 2026, calling for priority attention to be given to network security protection in the industry.
Affected by frequent cyber attacks on artificial intelligence models and intelligent agents, leading tech companies OpenAI, Anthropic, and Google jointly issued a statement on August 28, 2026, officially calling for priority attention to and enhanced cybersecurity measures in the industry.
科技 · AIacross 2 days
Three leading AI companies jointly issued a statement on August 28, 2026, calling for priority attention to be given to network security protection in the industry.
Affected by frequent cyber attacks on models and intelligent agents, the three companies jointly issued a statement, officially calling for priority attention to be given to and strengthening network security protection.
This feature improves video understanding accuracy by dynamically adjusting processing speed, reducing Token consumption by up to 88%.
The artificial intelligence industry has been affected by frequent network attacks on models and agents. OpenAI, Anthropic, and Google have jointly called for prioritizing cybersecurity issues.
Google has integrated the “active video understanding” function into its Gemini series of models. This feature aims to improve the accuracy of video content understanding and reduce analysis costs by dynamically adjusting processing speed. It enables AI to automatically determine viewing strategies, such as re-watching specific segments or switching between audio and visual sources, thereby addressing issues such as missing key actions due to traditional frame-by-frame sampling and overflowing context in long videos. Benchmark tests show that after enabling this function, token consumption was reduced by up to 88%, analysis costs were lowered by up to 66%, and accuracy improved by up to 7%. Combined with data sources like YouTube and Maps, this technology can real-time analyze game tactics, provide explanations, filter out invalid monitoring noise, and implement structured问答 retrieval for video content in the “Ask YouTube” function.