{"id":249810,"date":"2025-10-30T09:47:07","date_gmt":"2025-10-30T09:47:07","guid":{"rendered":"https:\/\/www.newsbeep.com\/ca\/249810\/"},"modified":"2025-10-30T09:47:07","modified_gmt":"2025-10-30T09:47:07","slug":"ai-and-public-disclosure-legal-implications-for-inventions-and-ip-new-technology","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/ca\/249810\/","title":{"rendered":"AI And Public Disclosure: Legal Implications For Inventions And IP &#8211; New Technology"},"content":{"rendered":"<p>Smart &amp; Biggar are most popular: <\/p>\n<p>Increasingly, scientific authors have been including disclaimers&#13;<br \/>\nin publications relating to the use of Large Language Models&#13;<br \/>\n(LLMs), such as ChatGPT, Grok, DeepSeek and CoPilot.1&#13;<br \/>\nThese disclaimers generally do not concern the substance of the&#13;<br \/>\npaper but rather note assistance in &#8220;proofreading&#8221; or&#13;<br \/>\n&#8220;increasing readability.&#8221; On its face, having an LLM&#13;<br \/>\nimprove a paper&#8217;s legibility for the public and other&#13;<br \/>\nscientists or engineers, particularly in subject areas full of&#13;<br \/>\njargon, appears to be a valuable use of this technology. However,&#13;<br \/>\nauthors should be wary of the potential legal implications of such&#13;<br \/>\nuse.<\/p>\n<p>Any disclosure to the public of an invention before the patent&#13;<br \/>\napplication&#8217;s filing date can jeopardize the validity of the&#13;<br \/>\nresulting patent.<\/p>\n<p>The terms of use for some of the most popular LLMs often make no&#13;<br \/>\nreference to confidentiality and may explicitly state that certain&#13;<br \/>\ninformation resulting from user input will be used to train the LLM&#13;<br \/>\nor for the company&#8217;s &#8220;own business&#13;<br \/>\npurposes.&#8221;2 Thus, the ability of companies that&#13;<br \/>\nprovide LLM services to use or disclose user inputs is often&#13;<br \/>\nunclear. Even if disclosure to an LLM does not itself constitute&#13;<br \/>\npublic disclosure, the company behind the LLM may nevertheless have&#13;<br \/>\naccess to the input and the ability to disclose it. Further, even&#13;<br \/>\nif the company behind the LLM may have no intention to access or&#13;<br \/>\ndisclose the inputted confidential information, a data leak could&#13;<br \/>\nstill inadvertently reveal this information to malicious actors.&#13;<br \/>\nThis vulnerability also applies to other non-AI tools, as any&#13;<br \/>\nsoftware, website or service which receives confidential&#13;<br \/>\ninformation from an innovator could store that information on their&#13;<br \/>\nnetworked systems.<\/p>\n<p>Some LLM providers attempt to address these risks through paid&#13;<br \/>\n&#8220;business&#8221; and &#8220;enterprise&#8221; subscriptions,&#13;<br \/>\nwhich typically promise more robust security features. For example,&#13;<br \/>\nsome premium LLM offerings exclude user inputs, outputs, and&#13;<br \/>\nmetadata from model training by default, while other offerings&#13;<br \/>\nallow users to opt-out of model training for free.3&#13;<br \/>\nAlthough these measures may reduce the risk of data leaks, they do&#13;<br \/>\nnot eliminate the fact that, if used by an inventor to proofread a&#13;<br \/>\nmanuscript, pre-publication information about inventions could&#13;<br \/>\nstill be transmitted to, and stored, on the LLM provider&#8217;s&#13;<br \/>\nservers. It is questionable, however, whether this would be&#13;<br \/>\nconsidered disclosure to the public.<\/p>\n<p>Other providers offer so-called &#8220;air-gapped&#8221; LLMs,&#13;<br \/>\nwhich are deployed entirely within a company&#8217;s or&#13;<br \/>\ngovernment&#8217;s own computing environment and disconnected from&#13;<br \/>\noutside servers. These solutions appear to provide a stronger&#13;<br \/>\nsafeguard, but they are generally available only under custom&#13;<br \/>\ncontracts with government or defense agencies and remain out of&#13;<br \/>\nreach for most innovators.<\/p>\n<p>Given the opaque functioning of LLMs, submitting a manuscript&#13;<br \/>\nthat contains confidential information or details of potential&#13;<br \/>\ninventions may amount to a public disclosure of that information or&#13;<br \/>\nthose inventions.<\/p>\n<p>While the topic of public disclosure may seem self-evident given&#13;<br \/>\nthat a manuscript is ultimately intended for publication, the&#13;<br \/>\neditorial and publication process associated with many journals can&#13;<br \/>\ntake years. Accordingly, if a manuscript is disclosed to an LLM,&#13;<br \/>\nand such disclosure is considered &#8220;public disclosure,&#8221; it&#13;<br \/>\ncould be fatal to the novelty of an invention.<\/p>\n<p>Public disclosure rules vary by country<\/p>\n<p>Novelty, as a fundamental requirement for patentability, demands&#13;<br \/>\nthat an invention not be disclosed to the public anywhere in the&#13;<br \/>\nworld before the relevant filing date. However, the extent to which&#13;<br \/>\ndisclosures made by inventors themselves affect an invention&#8217;s&#13;<br \/>\nnovelty varies widely across jurisdictions. For example, Canada and&#13;<br \/>\nthe United States each provide a relatively permissive twelve-month&#13;<br \/>\ngrace period, allowing inventor-originated disclosures, such as an&#13;<br \/>\nearly conference presentation or manuscript circulation, to be&#13;<br \/>\npermitted if a patent application is filed within one year. By&#13;<br \/>\ncontrast, the European Patent Office adopts a strict approach,&#13;<br \/>\nwhere almost any pre-filing disclosure destroys novelty, subject&#13;<br \/>\nonly to narrow exceptions for unauthorized disclosures or&#13;<br \/>\ndisclosures at recognized exhibitions. As well, China offers only a&#13;<br \/>\nsix-month grace period, and only in limited circumstances, such as&#13;<br \/>\ngovernment-recognized exhibitions, certain academic conferences, or&#13;<br \/>\nunauthorized disclosure.<\/p>\n<p>Until courts around the world address whether using LLMs&#13;<br \/>\nconstitutes public disclosure of an invention or new legislation&#13;<br \/>\nprovides clarity, this issue will remain uncertain. Further, as we&#13;<br \/>\nhave seen with other issues, results may vary across&#13;<br \/>\njurisdictions.<\/p>\n<p>Best practices<\/p>\n<p>Although there is presently no clear judicial or legislative&#13;<br \/>\nguidance relating to the use of LLMs and public disclosure of&#13;<br \/>\ninventions, inventors and applicants should take care to maintain&#13;<br \/>\nstrict confidentiality practices regarding potential inventions.&#13;<br \/>\nMore particularly, users of LLMs should only input information and&#13;<br \/>\ndata that are:<\/p>\n<p>&#13;<br \/>\nnot bound by a confidentiality agreement or duty from&#13;<br \/>\ndisclosing; and&#13;<br \/>\n&#13;<br \/>\nunrelated to a potential invention.&#13;<\/p>\n<p>For example, applicants and inventors should avoid inputting&#13;<br \/>\ninvention descriptions, test data, research notes and other data&#13;<br \/>\nrelated to a potential invention into an LLM. Any information that&#13;<br \/>\nwill be input into an LLM should be scrutinized for any content&#13;<br \/>\nrelated to the potential invention, especially in view of any other&#13;<br \/>\ninformation about the invention that was previously inputted into&#13;<br \/>\nthe LLM. This will help reduce the risk that the LLM will&#13;<br \/>\naccumulate a description of the potential invention over multiple&#13;<br \/>\nuses, either by the same or multiple users.<\/p>\n<p>Recognizing the reality of large corporate applicants and&#13;<br \/>\nindividual inventors, workplace policies regulating the use of LLMs&#13;<br \/>\nmay also form part of effective confidentiality strategies. These&#13;<br \/>\npolicies may include documentation requirements, multi-stage review&#13;<br \/>\nof LLM input prompts, and auditing of LLM and other AI tool usage.&#13;<br \/>\nSome policies may also require the cycling of LLMs and AI tools&#13;<br \/>\nused, to avoid providing too much potentially sensitive data to a&#13;<br \/>\nsingle tool. Applicants and inventors should also carefully&#13;<br \/>\nconsider the data policies of the LLMs and AI tools they use.<\/p>\n<p>In summary, for IP-minded researchers and inventors, the&#13;<br \/>\nrisk-reward balance of using LLMs in contexts involving potential&#13;<br \/>\npublic disclosure and patentability appears to favour either&#13;<br \/>\navoiding their use altogether or, more practically, strategically&#13;<br \/>\nlimiting such use to aspects that could not constitute public&#13;<br \/>\ndisclosure of potential inventions.<\/p>\n<p>Footnotes<\/p>\n<p>1. For example, see the &#8220;Declaration of generative&#13;<br \/>\nAI and AI-assisted technologies in the writing process&#8221; in the&#13;<br \/>\npaper: <a href=\"https:\/\/doi.org\/10.1016\/j.tifs.2024.104562\" target=\"_blank\" rel=\"nofollow noopener\">https:\/\/doi.org\/10.1016\/j.tifs.2024.104562<\/a>.<\/p>\n<p>2. For example, ChatGPT and OpenAI (see the subheading&#13;<br \/>\n&#8220;Services for individuals, such as ChatGPT, Codex, and&#13;<br \/>\nSora&#8221; in the article here: <a href=\"https:\/\/help.openai.com\/en\/articles\/5722486-how-your-data-is-used-to-improve-model-performance\" target=\"_blank\" rel=\"nofollow noopener\">https:\/\/help.openai.com\/en\/articles\/5722486-how-your-data-is-used-to-improve-model-performance<\/a>.&#13;<br \/>\nAlso, Anthropic (see the update: <a href=\"https:\/\/www.anthropic.com\/news\/updates-to-our-consumer-terms\" target=\"_blank\" rel=\"nofollow noopener\">https:\/\/www.anthropic.com\/news\/updates-to-our-consumer-terms<\/a>).&#13;<br \/>\nAlso, Gemini (see the guidance: <a href=\"https:\/\/cloud.google.com\/gemini\/docs\/discover\/data-governance\" target=\"_blank\" rel=\"nofollow noopener\">https:\/\/cloud.google.com\/gemini\/docs\/discover\/data-governance<\/a>).<\/p>\n<p>3. For example, ChatGPT and OpenAI (see the subheading&#13;<br \/>\n&#8220;Services for businesses, such as ChatGPT Business, ChatGPT&#13;<br \/>\nEnterprise, and our API Platform&#8221; in the article: <a href=\"https:\/\/help.openai.com\/en\/articles\/5722486-how-your-data-is-used-to-improve-model-performance\" target=\"_blank\" rel=\"nofollow noopener\">https:\/\/help.openai.com\/en\/articles\/5722486-how-your-data-is-used-to-improve-model-performance<\/a>.<\/p>\n<p>The preceding is intended as a timely update on Canadian&#13;<br \/>\nintellectual property and technology law. The content is&#13;<br \/>\ninformational only and does not constitute legal or professional&#13;<br \/>\nadvice. To obtain such advice, please communicate with our offices&#13;<br \/>\ndirectly.<\/p>\n","protected":false},"excerpt":{"rendered":"Smart &amp; Biggar are most popular: Increasingly, scientific authors have been including disclaimers&#13; in publications relating to the&hellip;\n","protected":false},"author":2,"featured_media":249811,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[49,48,61],"class_list":["post-249810","post","type-post","status-publish","format-standard","has-post-thumbnail","category-technology","tag-ca","tag-canada","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/249810","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/comments?post=249810"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/249810\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media\/249811"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media?parent=249810"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/categories?post=249810"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/tags?post=249810"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}