{"id":106,"date":"2023-05-02T09:46:44","date_gmt":"2023-05-02T13:46:44","guid":{"rendered":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/?p=106"},"modified":"2023-05-16T11:50:48","modified_gmt":"2023-05-16T15:50:48","slug":"the-ai-update-may-2-2023","status":"publish","type":"post","link":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/2023\/05\/02\/the-ai-update-may-2-2023\/","title":{"rendered":"The AI Update | May 2, 2023"},"content":{"rendered":"<p><img loading=\"lazy\" decoding=\"async\" class=\"alignleft wp-image-96 size-full\" src=\"http:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-content\/uploads\/sites\/63\/2023\/04\/DM-AI-Update-e1681141844877.png\" alt=\"\" width=\"150\" height=\"60\" \/><em>#HelloWorld. We originally thought this edition would focus on OpenAI\u2019s attempts to self-regulate GPT usage, but the European Union had other plans for us. This past Thursday, news broke of an agreement to add generative AI tools to the AI Act, the EU\u2019s centerpiece AI legislation. So today\u2019s issue starts there, before discussing OpenAI\u2019s and others\u2019 recent announcements regarding training data access and usage. Let\u2019s stay smart together. (<\/em><em><a href=\"mailto:AI-Update@duanemorris.com?subject=Subscribe%20to%20the%20mailing%20list%20&amp;body=Please%20add%20me%20to%20The%20AI%20Update%20list.\">Subscribe to the mailing list<\/a> to receive future issues).<\/em><\/p>\n<p><strong>The EU\u2019s Artificial Intelligence Act: <\/strong>The EU has been debating a proposed AI Act since <a href=\"https:\/\/digital-strategy.ec.europa.eu\/en\/policies\/european-approach-artificial-intelligence\" target=\"_blank\" rel=\"noopener\">2018<\/a>. In 2021, it published a <a href=\"https:\/\/digital-strategy.ec.europa.eu\/en\/library\/proposal-regulation-laying-down-harmonised-rules-artificial-intelligence\" target=\"_blank\" rel=\"noopener\">legislative framework<\/a> that would classify AI products into one of four categories: unacceptable risk (and therefore forbidden); high risk (and therefore subject to regular risk assessments, independent testing, transparency disclosures, and strict data governance requirements); limited risk; and minimal risk. But this approach was developed before so-called \u201cfoundation models\u201d\u2014LLMs like ChatGPT and image generators like DALL-E and MidJourney\u2014exploded into the public consciousness. So questions remained about whether the AI Act would be adjusted to accommodate this new reality.<\/p>\n<p><!--more--><\/p>\n<p>The answer is yes. Although the text of the amendments has not yet been released, <em>Euractiv <\/em>and other news sources monitoring the EU legislative process <a href=\"https:\/\/www.euractiv.com\/section\/artificial-intelligence\/news\/meps-seal-the-deal-on-artificial-intelligence-act\/\" target=\"_blank\" rel=\"noopener\">reported at the end of last week<\/a> that European Parliament members agreed to bring foundation models into the AI Act\u2019s framework. Many of these models are likely to be regulated as \u201chigh risk\u201d systems under the AI Act\u2019s proposed taxonomy. And the amended AI Act contemplates at least two special requirements specific to foundation models:<\/p>\n<ul>\n<li style=\"list-style-type: none\">\n<ul>\n<li>According to <em>The Verge<\/em>, model deployers would have to publicly disclose any \u201c<a href=\"https:\/\/www.theverge.com\/2023\/4\/28\/23702437\/eu-ai-act-disclose-copyright-training-data-report\" target=\"_blank\" rel=\"noopener\">copyrighted material<\/a>\u201d used to train their models. The exact level of specificity required is unknown, but this disclosure mandate (if ultimately adopted) may prove exceedingly difficult. It\u2019s no secret that OpenAI\u2019s, Google\u2019s, and other deployers\u2019 foundation models are trained on massive scrapes of the entire web, which presumably include countless instances of copyrighted works.<\/li>\n<li>Downstream app developers fine-tuning or building on top of foundation models could also be subject to the AI Act\u2019s heightened requirements, as part of a liability-shifting program. If these developers \u201c<a href=\"https:\/\/www.euractiv.com\/section\/artificial-intelligence\/news\/ai-act-meps-close-in-on-rules-for-general-purpose-ai-foundation-models\/\" target=\"_blank\" rel=\"noopener\">substantially modify an AI system<\/a>\u201d by, for instance, running fine-tuning training on new datasets, those actions could bring the downstream developers under the regulatory regime.<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>Much remains before the AI Act is finally enacted. A full vote of the European Parliament is expected mid-June. Then comes a so-called \u201ctrilogue\u201d process between the Parliament and two of the EU\u2019s other main arms (the Council of Ministers and the European Council) to reconcile differences and arrive at final language for adoption. Whatever the final version of the AI Act, it is not expected to go into effect before 2025.<\/p>\n<p><strong>OpenAI\u2019s Self-Regulation: <\/strong>While the EU debates wide-ranging regulations, model developers are not passively awaiting their fate. OpenAI has reportedly made changes to ChatGPT in Italy to address the personal privacy and age-gating concerns that caused Italian regulators to temporarily ban the product in <a href=\"https:\/\/blogs.duanemorris.com\/artificialintelligence\/2023\/04\/04\/the-ai-update-april-4-2023\/\" target=\"_blank\" rel=\"noopener\">early April<\/a>. According to statements sent to <a href=\"https:\/\/apnews.com\/article\/chatgpt-openai-data-privacy-italy-b9ab3d12f2b2cfe493237fd2b9675e21\" target=\"_blank\" rel=\"noopener\">the AP<\/a> and <a href=\"https:\/\/www.theverge.com\/2023\/4\/28\/23702883\/chatgpt-italy-ban-lifted-gpdp-data-protection-age-verification\" target=\"_blank\" rel=\"noopener\"><em>The Verge<\/em><\/a>, OpenAI deployed a <a href=\"https:\/\/share.hsforms.com\/1UPy6xqxZSEqTrGDh4ywo_g4sk30\" target=\"_blank\" rel=\"noopener\">new web form<\/a> that permits EU users to object to having their data used for model training; published a new <a href=\"https:\/\/help.openai.com\/en\/articles\/7842364-how-chatgpt-and-our-language-models-are-developed\" target=\"_blank\" rel=\"noopener\">help-center article<\/a> describing how ChatGPT collects and uses personal data; and added an age-verification step to its sign-up process.<\/p>\n<p>More generally, even outside the EU, OpenAI is adopting measures to give users more control over their data. Last week, OpenAI announced that users can now turn off their conversation history, <a href=\"https:\/\/openai.com\/blog\/new-ways-to-manage-your-data-in-chatgpt\" target=\"_blank\" rel=\"noopener\">effectively opting out of<\/a> having those conversations used to \u201ctrain and improve OpenAI models.\u201d By default, OpenAI uses conversations to update its models, but under its new approach would retain conversations only for 30 days and \u201creview them only when needed to monitor abuse, before permanently deleting.\u201d<\/p>\n<p><strong>What we\u2019re reading and following<\/strong>: With attention now increasingly focused on model training data, the next big question is <em>valuation<\/em>\u2014how much will data holders charge and how much will model developers be willing to pay for access to that data? In the past two weeks, <a href=\"https:\/\/www.nytimes.com\/2023\/04\/18\/technology\/reddit-ai-openai-google.html?smid=url-share\" target=\"_blank\" rel=\"noopener\">Reddit<\/a> and <a href=\"https:\/\/www.wired.com\/story\/stack-overflow-will-charge-ai-giants-for-training-data\/?utm_source=onsite-share&amp;utm_medium=email&amp;utm_campaign=onsite-share&amp;utm_brand=wired\" target=\"_blank\" rel=\"noopener\">Stack Overflow<\/a> announced they will not be providing AI training access to their data for free, at least for use in commercial applications. Unsurprisingly, these data owners have not publicly shared what such access would cost.\u00a0According to <em>Wired<\/em>, accessing 200 million tweets on Twitter today would run you <a href=\"https:\/\/www.wired.com\/story\/twitter-data-api-prices-out-nearly-everyone\/\" target=\"_blank\" rel=\"noopener\">$210,000 monthly<\/a>\u2014perhaps a useful benchmark for future negotiations in this space, and surely a sign that model training, already expensive, is about to get more so.<\/p>\n<p><strong>What <em>should <\/em>we be following?<\/strong> Have suggestions for legal topics to cover in future editions? Please send them to <a href=\"mailto:AI-Update@duanemorris.com\">AI-Update@duanemorris.com<\/a>. We\u2019d love to hear from you and continue the conversation.<\/p>\n<p><strong><em>Editor-in-Chief<\/em><\/strong><strong>: <\/strong><a href=\"mailto:agoranin@duanemorris.com\">Alex Goranin<\/a><\/p>\n<p><strong><em>Deputy Editors<\/em><\/strong><strong>:<\/strong> <a href=\"mailto:mcmousley@duanemorris.com\">Matt Mousley<\/a> and <a href=\"mailto:tmarandola@duanemorris.com\">Tyler Marandola<\/a><\/p>\n<p><em>If you were forwarded this newsletter, <\/em><a href=\"mailto:AI-Update@duanemorris.com?subject=Subscribe%20to%20the%20mailing%20list%20&amp;body=Please%20add%20me%20to%20The%20AI%20Update%20list.\"><em>subscribe to the mailing list<\/em><\/a><em> to receive future issues.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"<p>#HelloWorld. We originally thought this edition would focus on OpenAI\u2019s attempts to self-regulate GPT usage, but the European Union had other plans for us. This past Thursday, news broke of an agreement to add generative AI tools to the AI Act, the EU\u2019s centerpiece AI legislation. So today\u2019s issue starts there, before discussing OpenAI\u2019s and &hellip; <\/p>\n<p class=\"link-more\"><a href=\"https:\/\/blogs.duanemorris.com\/artificialintelligence\/2023\/05\/02\/the-ai-update-may-2-2023\/\" class=\"more-link\">Continue reading<span class=\"screen-reader-text\"> &#8220;The AI Update | May 2, 2023&#8221;<\/span><\/a><\/p>\n","protected":false},"author":6,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[15,19,16,13,17],"ppma_author":[5],"class_list":["post-106","post","type-post","status-publish","format-standard","hentry","category-general","tag-alex-goranin","tag-intellectual-property-litigation","tag-matt-mousley","tag-theaiupdate","tag-tylermarandola"],"authors":[{"term_id":5,"user_id":6,"is_guest":0,"slug":"duanemorris3","display_name":"Duane Morris","avatar_url":"https:\/\/secure.gravatar.com\/avatar\/843ff6e7a8fe5fc92109b47a45f34b6cf0ea499e6e788db23456c838b0ae6747?s=96&d=blank&r=g","0":null,"1":"","2":"","3":"","4":"","5":"","6":"","7":"","8":""}],"_links":{"self":[{"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/posts\/106","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/users\/6"}],"replies":[{"embeddable":true,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/comments?post=106"}],"version-history":[{"count":0,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/posts\/106\/revisions"}],"wp:attachment":[{"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/media?parent=106"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/categories?post=106"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/tags?post=106"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/ppma_author?post=106"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}