{"id":81534,"date":"2024-02-28T11:20:17","date_gmt":"2024-02-28T19:20:17","guid":{"rendered":"https:\/\/techreviewers.net\/what-is-sora-a-new-generative-ai-tool-could-transform-video-production-and-amplify-risks\/"},"modified":"2024-02-28T11:20:17","modified_gmt":"2024-02-28T19:20:17","slug":"what-is-sora-a-new-generative-ai-tool-could-transform-video-production-and-amplify-risks","status":"publish","type":"post","link":"https:\/\/techreviewers.net\/what-is-sora-a-new-generative-ai-tool-could-transform-video-production-and-amplify-risks\/","title":{"rendered":"What is Sora? A new generative AI tool could transform video production \u2014 and amplify risks"},"content":{"rendered":"<div id=\"bsf_rt_marker\"><\/div>\n<div id=\"article-main-content\">\n<p>This month, OpenAI announced a new <a href=\"https:\/\/openai.com\/sora\" rel=\"nofollow noopener\" target=\"_blank\">generative AI system named Sora<\/a>, which produces short videos from text prompts. While Sora is not yet available to the public, the high quality of the sample outputs published so far has provoked both <a href=\"https:\/\/www.aljazeera.com\/news\/2024\/2\/16\/out-of-this-world-openais-text-to-video-tool-sora-sets-internet-alight\" rel=\"nofollow noopener\" target=\"_blank\">excited<\/a> and <a href=\"https:\/\/www.newscientist.com\/article\/2417639-realism-of-openais-sora-video-generator-raises-security-concerns\/\" rel=\"nofollow noopener\" target=\"_blank\">concerned<\/a> reactions.<\/p>\n<p>The <a href=\"https:\/\/openai.com\/sora\" rel=\"nofollow noopener\" target=\"_blank\">sample videos<\/a> published by OpenAI, which the company says were created directly by Sora without modification, show outputs from prompts like \u201cphotorealistic closeup video of two pirate ships battling each other as they sail inside a cup of coffee\u201d and \u201chistorical footage of California during the gold rush\u201d.<\/p>\n<figure>\n<p><iframe allowfullscreen=\"allowfullscreen\" frameborder=\"0\" height=\"260\" srcdoc=\"\n\n<style>*{padding:0;margin:0;overflow:hidden}html,body{background:#000;height: 100%}img{position:absolute;top:0;left:0;width: 100%;height: 100%;object-fit:cover;transition:opacity .1s cubic-bezier(0.4,0,1,1)}a:hover img+img{opacity:1!important}<\/style>\n<p><a href='https:\/\/www.youtube.com\/embed\/HK6y8DAPN_0?feature=oembed&#038;autoplay=1&#038;mute=1&#038;modestbranding=1&#038;iv_load_policy=3&#038;theme=light&#038;playsinline=1'><img src='https:\/\/img.youtube.com\/vi\/HK6y8DAPN_0\/hqdefault.jpg'><img src='https:\/\/cdn0.tnwcdn.com\/wp-content\/themes\/cyberdelia\/assets\/img\/ytplaybtn.png' style='top: 50%;left:50%;width:68px;height:48px;transform:translate3d(-50%,-50%,0)'><img src='https:\/\/cdn0.tnwcdn.com\/wp-content\/themes\/cyberdelia\/assets\/img\/ytplaybtn-hover.png' style='top: 50%;left:50%;width:68px;height:48px;opacity:0;transform:translate3d(-50%,-50%,0)'><\/a>&#8221; width=&#8221;440&#8243;><\/iframe><\/p>\n<\/figure>\n<p>At first glance, it is often hard to tell they are generated by <a href=\"https:\/\/thenextweb.com\/topic\/artificial-intelligence\" rel=\"noopener\" target=\"_blank\">AI<\/a>, due to the high quality of the videos, textures, dynamics of scenes, camera movements, and a good level of consistency.<\/p>\n<p>OpenAI chief executive Sam Altman also posted some videos to X (formerly Twitter) generated in response to user-suggested prompts, to demonstrate Sora\u2019s capabilities.<\/p>\n<h2>How does Sora work?<\/h2>\n<div>\n<div id=\"hs-embed-tnw\">\n<p><img decoding=\"async\" src=\"https:\/\/s3.amazonaws.com\/events.tnw\/hardfork-2018\/uploads\/visuals\/tnw-newsletter.png\"><\/img><\/p>\n<div>\n<p>The <3 of EU tech<\/p>\n<p>The latest rumblings from the EU tech scene, a story from our wise ol&#8217; founder Boris, and some questionable AI art. It&#8217;s free, every week, in your inbox. Sign up now!<\/p>\n<\/div>\n<\/div>\n<\/div>\n<p>Sora combines features of text and image generating tools in what is called a \u201c<a href=\"https:\/\/openai.com\/research\/video-generation-models-as-world-simulators\" rel=\"nofollow noopener\" target=\"_blank\">diffusion transformer model<\/a>.\u201d<\/p>\n<p>Transformers are a type of neural network first <a href=\"https:\/\/dl.acm.org\/doi\/abs\/10.5555\/3295222.3295349\" rel=\"nofollow noopener\" target=\"_blank\">introduced by Google in 2017<\/a>. They are best known for their use in large language models such as ChatGPT and Google Gemini.<\/p>\n<p>Diffusion models, on the other hand, are the foundation of many AI image generators. They work by starting with random noise and iterating towards a \u201cclean\u201d image that fits an input prompt.<\/p>\n<figure><a href=\"https:\/\/images.theconversation.com\/files\/576444\/original\/file-20240219-30-a2zlmu.jpg?ixlib=rb-1.1.0&#038;q=45&#038;auto=format&#038;w=1000&#038;fit=clip\" rel=\"nofollow noopener\" target=\"_blank\"><img alt=\"A series of images showing a picture of a castle emerging from static.\" data-src=\"https:\/\/images.theconversation.com\/files\/576444\/original\/file-20240219-30-a2zlmu.jpg?ixlib=rb-1.1.0&#038;q=45&#038;auto=format&#038;w=754&#038;fit=clip\" data- decoding=\"async\"  src=\"data:image\/gif;base64,R0lGODlhAQABAIAAAAAAAP\/\/\/yH5BAEAAAAALAAAAAABAAEAAAIBRAA7\"><\/img><\/a><figcaption><span>Diffusion models (in this case Stable Diffusion) generate images from noise over many iterations.<\/span> <span><a href=\"https:\/\/en.wikipedia.org\/wiki\/Diffusion_model#\/media\/File:X-Y_plot_of_algorithmically-generated_AI_art_of_European-style_castle_in_Japan_demonstrating_DDIM_diffusion_steps.png\" rel=\"nofollow noopener\" target=\"_blank\">Stable Diffusion \/ Benlisquare \/ Wikimedia<\/a>, <a href=\"http:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/\" rel=\"nofollow noopener\" target=\"_blank\">CC BY-SA<\/a><\/span><\/figcaption><\/figure>\n<p>A video can be made from a sequence of such images. However, in a video, coherence and consistency between frames are essential.<\/p>\n<p>Sora uses the transformer architecture to handle how frames relate to one another. While transformers were initially designed to find patterns in tokens representing text, Sora instead uses tokens representing <a href=\"https:\/\/openai.com\/research\/video-generation-models-as-world-simulators\" rel=\"nofollow noopener\" target=\"_blank\">small patches of space and time<\/a>.<\/p>\n<h2>Leading the pack<\/h2>\n<p>Sora is not the first text-to-video model. Earlier models include <a href=\"https:\/\/ai.meta.com\/blog\/emu-text-to-video-generation-image-editing-research\/\" rel=\"nofollow noopener\" target=\"_blank\">Emu<\/a> by Meta, <a href=\"https:\/\/research.runwayml.com\/gen2\" rel=\"nofollow noopener\" target=\"_blank\">Gen-2<\/a> by Runway, <a href=\"https:\/\/stability.ai\/news\/stable-video-diffusion-open-ai-video-model\" rel=\"nofollow noopener\" target=\"_blank\">Stable Video Diffusion<\/a> by Stability AI, and recently <a href=\"https:\/\/lumiere-video.github.io\/\" rel=\"nofollow noopener\" target=\"_blank\">Lumiere<\/a> by Google.<\/p>\n<p>Lumiere, released just a few weeks ago, <a href=\"https:\/\/arxiv.org\/abs\/2401.12945\" rel=\"nofollow noopener\" target=\"_blank\">claimed<\/a> to produce better video than its predecessors. But Sora appears to be more powerful than Lumiere in at least some respects.<\/p>\n<p>Sora can generate videos with a resolution of up to 1920 \u00d7 1080 pixels, and in a variety of aspect ratios, while Lumiere is limited to 512 \u00d7 512 pixels. Lumiere\u2019s videos are around 5 seconds long, while Sora makes videos up to 60 seconds.<\/p>\n<p>Lumiere cannot make videos composed of multiple shots, while Sora can. Sora, like other models, is also reportedly capable of video-editing tasks such as creating videos from images or other videos, combining elements from different videos, and extending videos in time.<\/p>\n<figure>\n<p><iframe allowfullscreen=\"allowfullscreen\" frameborder=\"0\" height=\"260\" srcdoc=\"\n\n<style>*{padding:0;margin:0;overflow:hidden}html,body{background:#000;height: 100%}img{position:absolute;top:0;left:0;width: 100%;height: 100%;object-fit:cover;transition:opacity .1s cubic-bezier(0.4,0,1,1)}a:hover img+img{opacity:1!important}<\/style>\n<p><a href='https:\/\/www.youtube.com\/embed\/xNb7E_h48uw?feature=oembed&#038;autoplay=1&#038;mute=1&#038;modestbranding=1&#038;iv_load_policy=3&#038;theme=light&#038;playsinline=1'><img src='https:\/\/img.youtube.com\/vi\/xNb7E_h48uw\/hqdefault.jpg'><img src='https:\/\/cdn0.tnwcdn.com\/wp-content\/themes\/cyberdelia\/assets\/img\/ytplaybtn.png' style='top: 50%;left:50%;width:68px;height:48px;transform:translate3d(-50%,-50%,0)'><img src='https:\/\/cdn0.tnwcdn.com\/wp-content\/themes\/cyberdelia\/assets\/img\/ytplaybtn-hover.png' style='top: 50%;left:50%;width:68px;height:48px;opacity:0;transform:translate3d(-50%,-50%,0)'><\/a>&#8221; width=&#8221;440&#8243;><\/iframe><\/p>\n<\/figure>\n<p>Both models generate broadly realistic videos, but may suffer from hallucinations. Lumiere\u2019s videos may be more easily recognised as AI-generated. Sora\u2019s videos look more dynamic, having more interactions between elements.<\/p>\n<p>However, in many of the example videos inconsistencies become apparent on close inspection.<\/p>\n<h2>Promising applications<\/h2>\n<p>Video content is currently produced either by filming the real world or by using special effects, both of which can be costly and time consuming. If Sora becomes available at a reasonable price, people may start using it as a prototyping software to visualise ideas at a much lower cost.<\/p>\n<p>Based on what we know of Sora\u2019s capabilities it could even be used to create short videos for some applications in entertainment, advertising, and education.<\/p>\n<p>OpenAI\u2019s <a href=\"https:\/\/openai.com\/research\/video-generation-models-as-world-simulators\" rel=\"nofollow noopener\" target=\"_blank\">technical paper<\/a> about Sora is titled \u201cVideo generation models as world simulators.\u201d The paper argues that bigger versions of video generators like Sora may be, \u201ccapable simulators of the physical and digital world, and the objects, animals, and people that live within them.\u201d<\/p>\n<p>If this is correct, future versions may have scientific applications for physical, chemical, and even societal experiments. For example, one might be able to test the impact of tsunamis of different sizes on different kinds of infrastructure \u2014 and on the physical and mental health of the people nearby.<\/p>\n<p>Achieving this level of simulation is highly challenging, and some experts say a system like Sora is <a href=\"https:\/\/twitter.com\/fchollet\/status\/1758896780576739485\" rel=\"nofollow noopener\" target=\"_blank\">fundamentally incapable<\/a> of doing it.<\/p>\n<p>A complete simulator would need to calculate physical and chemical reactions at the most detailed levels of the universe. However, simulating a rough approximation of the world and making realistic videos to human eyes might be within reach in the coming years.<\/p>\n<h2>Risks and ethical concerns<\/h2>\n<p>The main concerns around tools like Sora revolve around their societal and ethical impact. In a world already <a href=\"https:\/\/www.who.int\/health-topics\/infodemic\/understanding-the-infodemic-and-misinformation-in-the-fight-against-covid-19#tab=tab_1\" rel=\"nofollow noopener\" target=\"_blank\">plagued by disinformation<\/a>, tools like Sora may make things worse.<\/p>\n<p>It\u2019s easy to see how the ability to generate realistic video of any scene you can describe could be used to spread convincing fake news or throw doubt on real footage. It may endanger public health measures, be used to influence elections, or even burden the justice system with <a href=\"https:\/\/www.jdsupra.com\/legalnews\/ai-threatens-courts-with-fake-evidence-7371356\/\" rel=\"nofollow noopener\" target=\"_blank\">potential fake evidence<\/a>.<\/p>\n<p>Video generators may also enable direct threats to targeted individuals, via deepfakes \u2014 particularly <a href=\"https:\/\/en.wikipedia.org\/wiki\/Deepfake_pornography\" rel=\"nofollow noopener\" target=\"_blank\">pornographic ones<\/a>. These may have terrible repercussions on the lives of the affected individuals and their families.<\/p>\n<p>Beyond these concerns, there are also questions of copyright and intellectual property. Generative AI tools require vast amounts of data for training, and OpenAI has not revealed where Sora\u2019s training data came from.<\/p>\n<p>Large language models and image generators have also been criticised for this reason. In the United States, a <a href=\"https:\/\/abcnews.go.com\/US\/famous-authors-lawsuit-chatgpt-maker-openai-begins-initial\/story?id=105239215\" rel=\"nofollow noopener\" target=\"_blank\">group of famous authors have sued OpenAI<\/a> over a potential misuse of their materials. The case argues that large language models and the companies who use them are stealing the authors\u2019 work to create new content.<\/p>\n<p>It is not the first time in recent memory that technology has run ahead of the law. For instance, the question of the obligations of social media platforms in moderating content has created heated debate in the past couple of years \u2014 much of it revolving around <a href=\"https:\/\/www.vox.com\/recode\/2020\/5\/28\/21273241\/section-230-explained-supreme-court-social-media\" rel=\"nofollow noopener\" target=\"_blank\">Section 230 of the US Code<\/a>.<\/p>\n<p>While these concerns are real, based on past experience we would not expect them to stop the development of video-generating technology. OpenAI <a href=\"https:\/\/openai.com\/sora#safety\" rel=\"nofollow noopener\" target=\"_blank\">says<\/a> it is \u201ctaking several important safety steps\u201d before making Sora available to the public, including working with experts in \u201cmisinformation, hateful content, and bias\u201d and \u201cbuilding tools to help detect misleading content.\u201d<\/p>\n<p><em><a href=\"https:\/\/theconversation.com\/profiles\/vahid-pooryousef-1508522\" rel=\"nofollow noopener\" target=\"_blank\">Vahid Pooryousef<\/a>, PhD candidate in Human Computer Interaction, <a href=\"https:\/\/theconversation.com\/institutions\/monash-university-1065\" rel=\"nofollow noopener\" target=\"_blank\">Monash University<\/a> and <a href=\"https:\/\/theconversation.com\/profiles\/lonni-besancon-591073\" rel=\"nofollow noopener\" target=\"_blank\">Lonni Besan\u00e7on<\/a>, Assistant Professor in Data Visualization, <a href=\"https:\/\/theconversation.com\/institutions\/linkoping-university-839\" rel=\"nofollow noopener\" target=\"_blank\">Link\u00f6ping University<\/a><\/em><\/p>\n<p><em>This article is republished from <a href=\"https:\/\/theconversation.com\" rel=\"nofollow noopener\" target=\"_blank\">The Conversation<\/a> under a Creative Commons license. Read the <a href=\"https:\/\/theconversation.com\/what-is-sora-a-new-generative-ai-tool-could-transform-video-production-and-amplify-disinformation-risks-223850\" rel=\"nofollow noopener\" target=\"_blank\">original article<\/a>.<\/em><\/p>\n<\/p><\/div>\n","protected":false},"excerpt":{"rendered":"<p>This month, OpenAI announced a new generative AI system named Sora, which produces short videos from text prompts. While Sora is not yet available to the public, the high quality of the sample outputs published so far has provoked both excited and concerned reactions. The sample videos published by OpenAI, [&hellip;]<\/p>\n","protected":false},"author":41,"featured_media":81535,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ocean_post_layout":"","ocean_both_sidebars_style":"","ocean_both_sidebars_content_width":0,"ocean_both_sidebars_sidebars_width":0,"ocean_sidebar":"","ocean_second_sidebar":"","ocean_disable_margins":"enable","ocean_add_body_class":"","ocean_shortcode_before_top_bar":"","ocean_shortcode_after_top_bar":"","ocean_shortcode_before_header":"","ocean_shortcode_after_header":"","ocean_has_shortcode":"","ocean_shortcode_after_title":"","ocean_shortcode_before_footer_widgets":"","ocean_shortcode_after_footer_widgets":"","ocean_shortcode_before_footer_bottom":"","ocean_shortcode_after_footer_bottom":"","ocean_display_top_bar":"default","ocean_display_header":"default","ocean_header_style":"","ocean_center_header_left_menu":"","ocean_custom_header_template":"","ocean_custom_logo":0,"ocean_custom_retina_logo":0,"ocean_custom_logo_max_width":0,"ocean_custom_logo_tablet_max_width":0,"ocean_custom_logo_mobile_max_width":0,"ocean_custom_logo_max_height":0,"ocean_custom_logo_tablet_max_height":0,"ocean_custom_logo_mobile_max_height":0,"ocean_header_custom_menu":"","ocean_menu_typo_font_family":"","ocean_menu_typo_font_subset":"","ocean_menu_typo_font_size":0,"ocean_menu_typo_font_size_tablet":0,"ocean_menu_typo_font_size_mobile":0,"ocean_menu_typo_font_size_unit":"px","ocean_menu_typo_font_weight":"","ocean_menu_typo_font_weight_tablet":"","ocean_menu_typo_font_weight_mobile":"","ocean_menu_typo_transform":"","ocean_menu_typo_transform_tablet":"","ocean_menu_typo_transform_mobile":"","ocean_menu_typo_line_height":0,"ocean_menu_typo_line_height_tablet":0,"ocean_menu_typo_line_height_mobile":0,"ocean_menu_typo_line_height_unit":"","ocean_menu_typo_spacing":0,"ocean_menu_typo_spacing_tablet":0,"ocean_menu_typo_spacing_mobile":0,"ocean_menu_typo_spacing_unit":"","ocean_menu_link_color":"","ocean_menu_link_color_hover":"","ocean_menu_link_color_active":"","ocean_menu_link_background":"","ocean_menu_link_hover_background":"","ocean_menu_link_active_background":"","ocean_menu_social_links_bg":"","ocean_menu_social_hover_links_bg":"","ocean_menu_social_links_color":"","ocean_menu_social_hover_links_color":"","ocean_disable_title":"default","ocean_disable_heading":"default","ocean_post_title":"","ocean_post_subheading":"","ocean_post_title_style":"","ocean_post_title_background_color":"","ocean_post_title_background":0,"ocean_post_title_bg_image_position":"","ocean_post_title_bg_image_attachment":"","ocean_post_title_bg_image_repeat":"","ocean_post_title_bg_image_size":"","ocean_post_title_height":0,"ocean_post_title_bg_overlay":0.5,"ocean_post_title_bg_overlay_color":"","ocean_disable_breadcrumbs":"default","ocean_breadcrumbs_color":"","ocean_breadcrumbs_separator_color":"","ocean_breadcrumbs_links_color":"","ocean_breadcrumbs_links_hover_color":"","ocean_display_footer_widgets":"default","ocean_display_footer_bottom":"default","ocean_custom_footer_template":"","ocean_post_oembed":"","ocean_post_self_hosted_media":"","ocean_post_video_embed":"","ocean_link_format":"","ocean_link_format_target":"self","ocean_quote_format":"","ocean_quote_format_link":"post","ocean_gallery_link_images":"on","ocean_gallery_id":[],"footnotes":""},"categories":[2668,1687,2666],"tags":[2667,1686,2665],"_links":{"self":[{"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/posts\/81534"}],"collection":[{"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/users\/41"}],"replies":[{"embeddable":true,"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/comments?post=81534"}],"version-history":[{"count":0,"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/posts\/81534\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/media\/81535"}],"wp:attachment":[{"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/media?parent=81534"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/categories?post=81534"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/techreviewers.net\/wp-json\/wp\/v2\/tags?post=81534"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}