![\"\"](\"http:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-content\/uploads\/sites\/63\/2023\/04\/DM-AI-Update-e1681141844877.png\")
<\/p>\n<\/p>\n
#HelloWorld. In this issue, we survey the tech industry\u2019s private self-regulation\u2014what model developers and online platforms have implemented as restrictions on AI usage. Also, one court in the U.S. hints at how much copying by an AI model is too much and the EU releases its most recent amendments to the AI Act. Let\u2019s stay smart together. (<\/em>Subscribe to the mailing list<\/a> to receive future issues).<\/em><\/p>\n Industry self-regulation: <\/strong>Past AI Updates<\/a> have summarized legislative and regulatory initiatives around the newest AI architectures\u2014LLMs and other \u201cfoundation models\u201d like GPT and Midjourney. In the meantime, LLM developers and users have not stood still. In our last issue, we discussed OpenAI\u2019s new user opt-out procedures<\/a>. While comprehensive private standards feel a long way off, here\u2019s what a few other industry players are doing:<\/p>\n <\/p>\n The cut-across theme for many other developers and platforms is restricting misleading and inauthentic AI content\u2014including generated AI output that is held out as human.<\/p>\n The Northern District of California Gives a Hint<\/strong>: A hot topic in the pending intellectual property lawsuits over generative AI is what level of copying of copyrighted material, at what stage of the AI development process (model training vs. output generation), may cross the line into infringement. The Northern District of California in the Doe 1 v. GitHub <\/em>class action just issued an opinion<\/a> that, while focused on procedural issues, left an intriguing clue: Showing that an AI model \u201cabout 1% of the time\u201d reproduces as output the copyrighted material it was trained on is enough of a \u201crealistic danger\u201d to permit claims seeking injunctive relief.<\/p>\n Compromise Amendments to the EU\u2019s AI Act<\/strong>: The EU continues to set the pace globally on AI regulation. In our last issue<\/a>, we summarized the history of the EU\u2019s efforts. The European Parliament has now released 144 pages of \u201ccompromise amendments<\/a>\u201d for further discussion among the three key legislative entities within the EU (the \u201ctrilogue\u201d!). It\u2019s a dense document, but the headline is that foundation models like GPT would be regulated on par with \u201chigh-risk AI systems.\u201d Want to dive in? Start with revised Articles 28, 28(a), 28(b), and 29, which provide a long list of \u201cresponsibilities\u201d to be imposed \u201calong the AI value chain\u201d from model developers through to model deployers.<\/p>\n What we\u2019re reading<\/strong>: How can you imperceptibly mark a piece of digital content as AI-generated? Digital watermarking of image files (essentially, careful manipulation of select pixel values) is well known. But researchers are also developing ways to watermark the generated language <\/em>output of LLMs<\/a> like GPT. It\u2019s known as \u201cstatistical watermarking.\u201d The core idea is to bias the model\u2019s selection of certain words to keep meaning the same while creating a word-use frequency outside of expected probabilities: Think text using the word \u201ccomprehend\u201d instead of \u201cunderstand\u201d more often than you would anticipate. This approach would work only for longer pieces of generated text, 800 words or more.<\/p>\n What should <\/em>we be following? <\/strong>Have suggestions for legal topics to cover in future editions? Please send them to AI-Update@duanemorris.com<\/a>. We\u2019d love to hear from you and continue the conversation.<\/p>\n Editor-in-Chief<\/em><\/strong>: <\/strong>Alex Goranin<\/a><\/p>\n Deputy Editors<\/em><\/strong>:<\/strong> Matt Mousley<\/a> and Tyler Marandola<\/a><\/p>\n \u00a0<\/em>If you were forwarded this newsletter, <\/em>subscribe to the mailing list<\/em><\/a> to receive future issues.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":" #HelloWorld. In this issue, we survey the tech industry\u2019s private self-regulation\u2014what model developers and online platforms have implemented as restrictions on AI usage. Also, one court in the U.S. hints at how much copying by an AI model is too much and the EU releases its most recent amendments to the AI Act. Let\u2019s stay … <\/p>\n Continue reading “The AI Update | May 16, 2023”<\/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-108","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:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-content\/uploads\/sites\/63\/2020\/10\/dmlogo.jpg","author_category":"1","last_name":"Davies","first_name":"Thomas","job_title":"","user_url":"https:\/\/www.duanemorris.com","description":"Visit the Duane Morris website.<\/a>"}],"_links":{"self":[{"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/posts\/108","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=108"}],"version-history":[{"count":0,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/posts\/108\/revisions"}],"wp:attachment":[{"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/media?parent=108"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/categories?post=108"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/tags?post=108"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/blogs.duanemorris.com\/artificialintelligence\/wp-json\/wp\/v2\/ppma_author?post=108"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}\n
\n
\n
\n