Adobe CEO Shantanu Narayen on developing GenAI standards, building AI models, more

Adobe's approach to generative AI is built on more than a decade of experience integrating AI into the company's products, says CEO Shantanu Narayen

Adobe is collaborating with its customers, partners and other companies to develop broad industry standards for generative artificial intelligence (AI), said chief executive Shantanu Narayen.

The move comes amid growing worries over the potential misuse and abuse of this nascent technology.

"Responsible innovation cannot be an afterthought to all of the advancements that we make. We're working to combat misinformation by advocating for widespread adoption of Content Credentials as an industry standard for establishing trust in all of the digital content that's being created" Narayen said at Adobe Summit, the company's digital experience conference, on March 26.

Content Credentials is an open technology that allows creators to attach additional information or metadata to a piece of content, like images, to increase transparency about its origins and history. This helps people assess the credibility of the content they see online.

It was developed by Content Authenticity Initiative (CAI), a community of over 2,500 members comprising media and tech companies, NGOs, academics, and others. The community was started by Adobe, Twitter (now X) and New York Times in 2019.

The technology is based on an open technical specification developed and maintained by the C2PA (Coalition for Content Provenance and Authenticity), a cross-industry open standards development consortium co-founded by Adobe and a group of influential technology and media companies such as Microsoft, Intel, Arm, BBC and Truepic in 2021.

Google, Sony and Publicis Groupe have subsequently joined the steering committee of the global standards body.

Also read: Adobe, others join White House's voluntary commitments on AI

Over the past year or so, several tech companies including OpenAI and Facebook parent Meta have announced plans to implement Content Credentials in their respective products. Meta will be labelling AI-generated photos on Facebook, Instagram, and Threads through this technology while OpenAI will be implementing it for images generated in ChatGPT and through its API offering.

Camera maker Leica launched a camera with built-in Content Credentials in December 2023 while Nikon and Sony have announced plans to adopt it into their upcoming camera models. Chipmaker Qualcomm also added support for the technology in Snapdragon 8 Gen3 mobile platform that was announced in October 2023.

Last month, at the Munich Security Conference, over 20 leading technology companies including Adobe, Amazon, Google, IBM, Meta, Microsoft, OpenAI, Snap Inc, and X have signed a pact, pledging to work together to detect and counter harmful AI content.

The agreement, called Tech Accord to Combat Deceptive Use of AI in 2024 Elections, is a set of eight commitments to deploy technology countering harmful AI-generated content meant to deceive voters.

During the conference, Adobe also revealed an expanded integration of Content Credentials into the company's Creative and Enterprise applications including attaching this label to all the outputs generated by its family of creative generative AI models Adobe Firefly and adding support for it in assets generated in the company's content management system Adobe Experience Manager.

Read: Adobe unveils new GenAI tools to help marketers deliver more personalised experiences

Adobe's generative AI approach

In his address, Narayen said that Adobe's approach to generative AI is built on more than a decade of experience integrating AI into the company's products.

"To us, it's about innovating across the framework of data, models, and applications. Because AI is only as good as the data on which it's trained, our rich data draws upon the investments as well as the expertise across creativity, documents, and customer experience management" he said.

Narayen said that Firefly is trained purely on licensed and public domain content to ensure that it offers customers a commercially safe solution that doesn't infringe on any third party intellectual property rights.

He mentioned that Adobe will build high-quality foundation AI models in categories where they have deep domain expertise. The software giant will also strike strategic partnerships with other AI models to address a range of customer use cases "which are strengthened with our intellectual property and ensure that they have the privacy and governance protocols," Narayen said.

The Adobe chairman said the exponential value of AI will come from how successfully the company can integrate it into their daily tasks and workflows. To facilitate this, the software firm has integrated Firefly across the company's Creative Cloud, Document Cloud, and Experience Cloud applications. Firefly has generated over 6.5 billion images to date, he said.

"With AI, we are able to optimize the entire content-to-activation process, empower teams to work in real time, and produce what was once weeks or months long ideation into a matter of mere minutes" Narayen said.

Helping brands in their GenAI journey

During his keynote address, the Adobe chief also announced Firefly Services that enables brands to create their own custom AI models along with providing access to over 20 generative and creative APIs, tools and services that aims to eliminate manual and repetitive tasks currently performed by creative teams.

A new Adobe Express mobile app, which was introduced earlier this month, also brings Firefly's capabilities to mobile workflows, enabling anyone from a K-12 student to a marketer to quickly design and turn their ideas into content, modify assets and create standout designs, he said.

In February, Adobe also added an AI assistant in beta to its document management apps Reader and Acrobat, that can generate an instant summary of the document, answer questions through a conversational interface and facilitates people to generate emails, reports and presentations.

"So much of the world's information is embedded in PDF...This combines the power of generative AI with our unique understanding of the PDF format so that we can transform the way people interact with as well as extract additional value from their most important documents" Narayen said.

At the event, Adobe also unveiled Adobe GenStudio, a new generative AI application that enables brands to plan, create, store, deliver marketing content and measure the performance of their campaigns at a single place. The application, which was first previewed in September 2023, is expected to launch in the later part of Q3-FY24.

"One of the greatest opportunities we now have with AI is to actually personalise digital experiences to the exact individual rather than a segment or a cohort of people" Narayen said.

He mentioned that AI will empower businesses to rethink the customer engagement platforms and tackle them in more efficient ways.

"While the world has been captivated over the last year by generative AI, the reality is that every disruptive technology and every era of technology from the PC, Web, and mobile, cloud computing builds on the platforms of the past. With every era, Adobe has tried to invest, to transform each of the experiences that you can deliver across the entire customer's end-to-end journey with your brands" Narayen said.

(The reporter was in Las Vegas at the invitation of Adobe)