At Pinterest, for instance, we’re deploying AI to flip the script on social media, using it to more aggressively promote user well being rather than the alternative formula of triggering engagement by enragement. I believe AI can benefit our 600 million users for years to come and at a fraction the cost that many associate with the technology.

You don’t need to be a company with the size and scale of a Silicon Valley juggernaut to unlock the enormous economic benefits and innovation that AI could deliver. In fact, I have long argued that AI can be far more democratized than it currently is, leading to an entirely new generation of entrepreneurial success stories.

To realize AI’s real potential, however, commonly held assumptions need to be challenged. This is the only way to address a widespread public trust problem that AI currently faces and create an AI ecosystem that leaves no one behind.

For starters, our fundamental view of access to essential tools must change. Although often viewed as a battle between the largest (and often costly) proprietary models, a community of accessible open-source models is thriving in 2026, leveling the playing field for entrepreneurs eager to usher in the next generation of innovation.. 

Additionally, creators and publishers are no longer powerless when it comes to protecting the value of their work from companies that attempt to use their information to train generative AI models. The Napster phase of AI needs to end – now is the time for a clear exchange of value that benefits content creators.

Finally, regulation can no longer be viewed as the enemy. Oversight protects users and provides an incentive for private companies to compete on safe and positive user experiences. 

Open Source: Blueprint for Next Batch of Big Companies  

To date, too much of the focus of AI is who is building the largest proprietary models. The race to build powerful models is profound, but the discussion needs to include far more emphasis on open source and its ability to spur innovation across the broad business community.  

Pinterest recently announced an important milestone that demonstrates the potential. In our quest to harness the power of AI, we were able to tap into available large-scale open-source models and achieve performance similar to proprietary models but at 90% less cost. This addresses the return-on-AI-investment headache that many CEOs are facing as they spend a fortune on off-the-shelf proprietary solutions that don’t yet offer commensurate savings. 

This is not a new phenomenon. For decades, open-source software has been an important accelerant to nascent industries. Many of today’s largest companies, including trillion-dollar market cap enterprises, wouldn’t exist if they had to rely on proprietary databases or operating systems. 

The next batch of world-changing companies should follow a similar blueprint. Otherwise, we risk seeing the proprietary software companies collect all the value and, as a result, stifle innovation and ultimately block AI’s long-term potential. 

• Ownership Matters – Ditching AI’s Napster Era

Social media platforms like Pinterest live and die by users’ willingness to share creative and original ideas. 

Thankfully, humans feed the internet with a trove of new information every day. That information is inspired and validated by a level of creativity, reasoning, and work ethic that even the most advanced generative AI models do not possess. 

This presents large language models with a daunting learning curve that cannot be addressed without real-time access to this torrent of fresh ideas. That access should not be unfettered. 

When AI disregards ownership, content creators are less inclined to share their work and the public discourse suffers. When AI respects ownership, these originators can thrive, and the public gets better information. 

Currently, AI’s approach has resembled the old Napster pirating model – where music could be downloaded by tens of millions of internet users at no cost – than the iTunes or Spotify model – where publishers get compensated every time their work is accessed. 

The good news is there are several frameworks emerging that solve this problem. One is Cloudflare’s new model that allows content creators to choose whether and how GenAI companies use their content. Cloudflare’s tool works as a pay-per-crawl service, distinguishing GenAI crawlers that take information without sending much traffic back to creators, from search crawlers, which actually drive traffic back to the original source. 

Supporting Regulation that Protects and Promotes

As hard as it is to believe, standard installation of seatbelts were once considered “bad for business” by carmakers. That changed when crash test ratings created an incentive to do the right thing. The same can happen with standards in tech that protect users and promote responsible innovation. 

If you spend much time playing with AI, you’ll understand the need for regulations to stop a race to the bottom currently influencing the industry. No company should allow chatbots to have sexually explicit conversations with children, for instance. People, meanwhile, should be protected from bad actors attempting to use AI to manipulate their images or other information.

The question is what does meaningful regulation look like?

The App Store Accountability Act is one example of where immediate progress could be made. By making the app store a one-stop shop for age verification and parental consent, we can create consistent protections from the very moment a device is first turned on. 

Additionally, Pinterest envisions a world where social media companies and AI companies are competing on their safety records. To do this, the industry will need baseline regulations that focus on outcomes and leave space for companies to innovate in how to exceed those basic expectations.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.