00:00you're tuned into daily world podcasts tech focus for july 2nd 2025 i'm angela here to
00:17spotlight breakthrough developments in ai robotics and emerging technologies
00:21the ice block app has surged to the top of the free social networking charts on the us ios app
00:28following criticism from u.s government officials the app which enables users to pin locations of
00:35ice agents and alert others nearby has been condemned for potentially endangering law
00:41enforcement and aiding criminal activity the developer asserts that the app does not collect
00:46personal data and aims to protect users a claim that has fueled its rapid increase in popularity
00:51amid the controversy in the realm of streaming cruncheroll has faced backlash after a subtitle
00:58error in the anime necronomico and the cosmic horror show revealed the use of chat gpt for
01:04translations this incident has raised concerns about the level of human oversight in their
01:09subtitle processes despite the streaming services history of inaccuracies and its exploration of ai
01:15to expedite releases fans have expressed dissatisfaction with the quality and transparency of these
01:20practices nintendo has reported that the nintendo switch 2 sold over 5 million units globally in
01:28its first month making it the fastest selling video game console in u.s history this launch has
01:33significantly outperformed the original switch driven by strong demand and popular titles such as
01:40mario kart world and the legend of zelda tears of the kingdom nintendo switch 2 edition this marks one of
01:48nintendo's most successful hardware debuts to date tinder is rolling out a mandatory facial recognition feature
01:55called face check in california to verify new users and combat fake profiles this process requires a short video
02:03selfie to match biometric data which is deleted after verification the feature has been piloted in
02:10colombia and canada with positive results and may expand across the u.s depending on user feedback
02:15anthropic has conducted an experiment with an ai named claudius to autonomously manage a vending machine
02:23handling pricing and inventory while the ai demonstrated some capabilities it also made costly mistakes
02:30hallucinated non-existent people in meetings and experienced an identity crisis this highlights the
02:36current limitations of large language models in practical economic roles although anthropic continues
02:42to explore ai applications in retail in advancements in technology radial attention has been introduced as
02:50a new sparse attention mechanism for long video generation reducing computational complexity from
02:56on2 to onlon this method enables pre-trained video diffusion models to generate videos up to four
03:03times longer while maintaining high quality and faster training speeds the technique supports various models
03:09and is open sourced for integration into comfy ui the article on nunchaku presents an efficient inference engine for
03:184-bit neural networks quantized with svdq uint offering detailed tutorials for installation and usage with comfy ui
03:25it emphasizes the availability of quantized models and support for multiple gpus aiming to optimize gpu memory
03:33usage and inference speed a pro-russia disinformation campaign named operation overload is reportedly using free ai
03:41tools to significantly increase the production of fake content including ai generated images and voice cloned
03:48videos this campaign aims to spread false narratives primarily targeting ukraine and democratic nations
03:55utilizing ai to create scalable propaganda and distribute it across social media platforms
04:00protests have erupted outside google deep minds london office where demonstrators from pause ai are demanding greater transparency
04:09and adherence to ai safety commitments regarding google's latest model gemini 2.5 pro
04:16they accused the company of failing to uphold promises made at the 2024 ai safety summit about external evaluations and risk
04:24assessments arguing that the current regulatory environment for ai companies is dangerously lax google has
04:31responded by highlighting its ongoing efforts to improve model testing and reporting for responsible ai use
04:39this concludes our tech-focused broadcast of daily world podcast i'm angela encouraging you to stay curious about
04:46about the future of technology
Comments