Protecting your Content in the Age of AI

By Paul Melcher, republished from Kaptur



A week ago, leading tech companies working on generative AI pledged to use their technical abilities to clearly mark content produced by their AI. While this move is a positive step, it only lightly touches on the broader problems of reliability and trust in the era of AI.

Generative AI poses a major threat to visual content. The issue isn’t simply that fake images could be mistaken for real ones. It’s actually more concerning that real images could lose their credibility. The former issue can lead to deception, while the latter can foster distrust. Distrust is particularly problematic because, unlike deception, which impacts a single event and can be rectified, distrust undermines the very foundation of a relationship and is essentially irreversible. As a result, none of your images, real or artificial, can be trusted.

In a world where everyone, from teenagers to Fortune 50 companies, communicates through images, this problem is crucial. It threatens not only our trust in the news but also our relationships with all kinds of brands.

Today, anyone can manipulate a corporate or marketing image into a tool for spreading misinformation with just a few AI-assisted clicks, leaving companies defenseless. Public companies are at the most significant risk of experiencing profound damages, as their stock market value is heavily influenced by shareholders’ perceptions. Tampering with existing product images or images of the CEO will soon become a significant means of negatively impacting their financial performance. We’ve proposed that Digital Asset Management (DAM) vendors adjust their tools to help companies build databases of verifiable truths, but there are other potential solutions for protecting visual content.

Let’s look at some pros and cons of various methods to protect visual content from malicious manipulation:

1. C2PA/CAI (Content Authenticity Initiative and Coalition for Content Provenance and Authenticity)

Pros:

• Creates a standard framework for attribution, making it easier to identify original creators.
• Works with existing metadata standards and can work with other technologies, like blockchain or watermarking, to further enhance content authenticity
• This initiative was created by a diverse and influential consortium, including tech giants like Adobe, Microsoft, Intel, Sony, media companies like BBC and New York Times, and camera manufacturers like Nikon and Canon, indicating strong industry support.
• It provides a way for users to understand the full history of a digital asset and verifies its origin via an approved certificate.
• Can handle all types of digital files (text, sound, video), increasing its applicability.

Cons:

• Young solution: Successful adoption requires worldwide consensus among various players in the tech and media industry.
• The process for creating approved certificates is not yet well-defined.
• It hasn’t been integrated into any image editing and management tools beyond some Adobe products like Photoshop.
• It’s still in early stages of development, thus requiring significant software development resources.

2. Crypto/Blockchain

Pros:

• Ensures content integrity by storing and verifying information about the digital asset in a decentralized manner.
• Content creators can maintain ownership rights of their work, and the chain of custody can be easily tracked.
• Can work in conjunction with initiatives like C2PA/CAI to further enhance content integrity.

Cons:

• High energy usage and carbon footprint associated with blockchain transactions.
• Involves technical complexity and can be expensive to implement.
• No industry standard in place currently, which can lead to interoperability issues.

Oct. 24 (pm + reception)

Oct. 25 (am + pm + reception)

San Francisco

Save $50 and buy your Early Bird Ticket now!

Show & Tell Demo Presenters Hall of Fame

2022 Visual 1st Awards winners:

Best of Show – Kandao

Best Technology – Twelve Labs

Best Business Potential – Zenfolio

Special Recognition – Zingcam

2021 Visual 1st Awards winners:

Best of Show – zerolens

Best Technology – autoRetouch (acquired)

Best Business Potential – Bloom AI

Special Recognition – vAIsual

2020 Visual 1st Awards winners:

Best of Show – Profoto

(gone public)

Best Technology – Dotphoton 

Best Business Potential – Mobius Labs

Special Recognition – Snapbar

2019 Visual 1st Awards winners:

Best of Show – Bellus3D

Best Technology – Skylum

Best Business Potential – GotPhoto (acquired)

Special Recognition – Generated Photos

2018 Visual 1st Awards winners:

Best of Show – Insta360

Best Technology – Embody Digital 

Best Business Potential – Capturelife

(acquired)

Special Recognition – Pixsy 

2017 Visual 1st Awards winners:

Best of Show – Photomyne

(gone public)

Best Technology – PastBook

Best Business Potential – V360

Special Recognition – Streetography

2016 Visual 1st Awards winners:

Best of Show – Funimate

Best Technology – Regaind (acquired)

Best Business Potential – Beamr (gone public)

Special Recognition – Keepy

[Click on text links to read the jury's perspectives on that year's winners]

View complete list of past presenters and panelists acquired since their appearance at Visual 1st.

Interested in doing a Show & Tell demo this year at no charge for attendees? Here's more info.

3. Watermarking (invisible)

Pros:

• Easy to implement and helps directly associate content with its original and creator.
• Can be used as proof of ownership.
• Can help identify any modifications by comparing with the original file.
• Can work in association with C2PA, Blockchain, and ISCC to ensure authenticity.

Cons:

• Requires linkage to an original public reference database for verification.
• Relies on proprietary certification tied to specific vendors, limiting versatility.

4. ISCC (International Standard Content Code)

Pros:

• Provides a unique identifier for digital content, aiding in content tracking and rights management.
• Facilitates detection of duplicate or very similar content.
• The ISCC is on its way to becoming an ISO standard, which could increase its adoption and influence.
• Can handle all types of digital files (text, sound, video), increasing its applicability. 

Cons:

• Might not prevent the initial misuse of content; it mainly helps in the aftermath.
• Widespread adoption requires consensus among stakeholders.
• No existing practical implementations besides the Liccium app, indicating it may take time to realize its full potential.

5. Image Matching

Pros:

• Can identify identical or near-identical images used in different contexts.
• Can work in association with C2PA, Blockchain, Watermarking and ISCC to ensure authenticity.
• Useful for tracking the spread and usage of an image.

Cons:

• Struggles to identify manipulated or heavily altered versions of an image.
• Can yield false positives, identifying unrelated images as matches.

6. AI

Pros:

• Can be trained to detect subtle manipulations in images that may be invisible to the human eye.
• Can automate the process of identifying and flagging manipulated content.

Cons:

• AI systems can be fooled or evaded by sophisticated manipulations.
• Requires significant computational resources and technical expertise to implement effectively.
• As of now, there are no viable AI solutions in the marketplace that effectively address the issue of image integrity.

None of these solutions provide a perfect out-of-the-box fix, and their appropriateness depends on specific requirements. Some are explicitly designed to address authenticity and trust issues, such as C2PA, but lack adoption, while others, like watermarking, are long-existing technologies but lack task-specific implementation.

However, the issue of image manipulation and consequent erosion of trust is only going to intensify, making the implementation of a solution more urgent. Companies must act now to protect their content integrity and build response mechanisms to mitigate the escalating impact of image manipulation and the erosion of trust in visual content.

Author: Paul Melcher

Paul Melcher is a highly influential and visionary leader in visual tech, with 20+ years of experience in licensing, tech innovation, and entrepreneurship.

He is the Managing Director of MelcherSystem and has held executive roles at Corbis, Stipple, and more. Melcher received a Digital Media Licensing Association Award and is a board member of Plus Coalition, Clippn, and Anthology, and has been named among the “100 most influential individuals in American photography.”



And a few more things...

Google Photos. Reliving your Memories 2.0. Google Photos introduces a new Memories view feature, building upon its four-year-old Memories feature aimed at recalling notable moments from your past.

The new version gives users more control to view their visual memories: they can now modify, rename or manually create Memories, and – of course these days – use generative AI to come up with title ideas. According to Google, the existing Memories feature is used by more than half a billion people each month.

I look forward to delving more into Google Photos’ strategic innovation considerations during our fireside chat session at Visual 1st with John Fisher, Sr. Director of Engineering of Google Photos.

Retouch4Me. AI-fueled image recoloring. Want your photo to be recolored with a color scheme similar to that of another photo you really like? Retouch4Me just announced a new AI plugin called Color Match that can be used to apply colors from a reference image to the photo that you're editing. 

Atmosfy. Yelp meets TikTok. How to best convey what a local restaurant, hotel or any other business looks and feels like? Atmosfy’s answer: short-form video. To post content to Atmosfy, users can record short videos of their experiences, add the business location and then upload them to the app. Once you have uploaded the video, you can add also tags and descriptions.

Founded in 2021, Atmosfy has shared video content in more than 10,000 cities across more than 150 countries. The company now announced it has raised $12M in seed funding led by Redpoint Ventures.

Content organizations call for unified AI regulation & practices. In an open letter 10 organizations, ranging from news agencies such as Agence France-Presse, to content creator organizations like the National Press Photographers Association, plus stock agency Getty Images, express their concern if the current AI advancements are left unchecked: 

“We, the undersigned organizations, support the responsible advancement and deployment of generative AI technology, while believing that a legal framework must be developed to protect the content that powers AI applications as well as maintain public trust in the media that promotes facts and fuels our democracies.” 

For more about the digital rights implications of generative AI, please check out the program of our partner, the IDMLC conference (special ticket options for Visual 1st attendees here); to hear the perspectives from stock photography executives from Shutterstock, Adobe Stock and Freepik on the responsible use of AI, join our “The future of Stock Photography ain’t what it used to be” panel session at Visual 1st.

Canto & MerlinOne. Consolidation in DAM land. DAM vendor Canto acquires fellow DAM vendor MerlinOne. According to Canto, with its acquisition of MerlinOne advanced AI and visual search capabilities will make organizing, finding, and distributing brand assets faster than ever before, even if the assets have scarce or no textual metadata.

Professional photography a dying profession? Nope – at least not in Holland. In the category of “I did not already think so”: The number of registered pro photographers in Holland has increased by 34% since 2018. Almost 60% of them say they can make a living with photography as their main source of income. While only 46% were women in 2018, their share is now above 50%. Brush up on your Dutch and read more here.

Visual 1st Keynote, Fireside Chat, and Panel topics to date:

Profiting from Major Technology Transformations – how about AI?

Google Photos – the challenge of “Anything is Possible”

TimMcCain: From homeless to success as a photo entrepreneur – and beyond

AI Image Creation – exploring where boundless creativity could take us

The future of Stock Photography ain’t what it used to be

What’s next for the Photo Print Market – today’s consumers are digital, online, and do video. Now what?

AI Image Curation 2.0 – when AI-based auto-tagging is so yesterday

Check out our latest Visual 1st Flash Viewpoint video with:

Rusty Pepper, Print-On-Demand Strategist, at Taylor Corporation

2 questions, 3 minutes:

- Why are Print on Demand solution providers expanding into the area of Photo Print Products?

- What about the other way around? Why are more Photo Print Product solution providers now adding non-photo Print on Demand products?

Check out previous Flash Viewpoints from thought leaders at VIEWBUG, Format, Edge Imaging, Picmentum, RCKIVE, MailPix and Mixbook.

Visual 1st 2023 Vendor Sponsors – 18 sponsors to date!