AI Developers Turn to Synthetic Data as Original Content Dries Up

As AI fashions eat the web’s free content material, a looming disaster is rising: What occurs when there’s nothing left to practice on?

A latest Copyleaks report revealed that DeepSeek, a Chinese AI mannequin, typically produces responses practically an identical to ChatGPT, elevating considerations that it was skilled on OpenAI outputs.

That’s led some to suspect the period of “low-hanging fruit” in AI growth could also be over.

In December, Google CEO Sundar Pichai acknowledged this actuality, warning that AI builders are quickly exhausting the availability of freely obtainable, high-quality coaching information.

“In the current generation of LLM models, roughly a few companies have converged at the top, but I think we’re all working on our next versions too,” Pichai stated on the New York Times’ annual Dealbook Summit in December. “I think the progress is going to get harder.”

With the availability of high-quality coaching information dwindling, many AI researchers are turning to artificial information generated by different AI.

Synthetic information isn’t new—it dates again to the late Nineteen Sixties—and has been utilized in statistics and machine studying, counting on algorithms and simulations to create synthetic datasets that mimic real-world data. But its rising function in AI growth sparks recent considerations, significantly as AI techniques combine into decentralized applied sciences.

Bootstrapping AI

“Synthetic data has been around in statistics forever—it’s called bootstrapping,” Professor of Software Engineering at MIT Muriel Médard advised Decrypt in an interview at ETH Denver 2025. “You start with actual data and think, ‘I want more but don’t want to pay for it. I’ll make it up based on what I have.’”

Medard, the co-founder of decentralized reminiscence infrastructure platform Optimum, stated the principle problem in coaching AI fashions isn’t the dearth of information however relatively its accessibility.

“You either search for more or fake it with what you have,” she stated. “Accessing data—especially on-chain, where retrieval and updates are crucial—adds another layer of complexity.”

AI builders face mounting privateness restrictions and restricted entry to real-world datasets, with artificial information turning into a vital different for mannequin coaching.

“As privacy restrictions and general content policies are backed with more and more protection, utilizing synthetic data will become a necessity, both out of ease of access and fear of legal recourse,” Senior Solutions Architect at Druid AI Nick Sanchez advised Decrypt.

“Currently, it’s not a perfect solution, as synthetic data can contain the same biases you would find in real-world data, but its role in handling consent, copyright, and privacy issues will only grow over time,” he added.

Risks and rewards

As using artificial information grows, so do considerations about its potential for manipulation and misuse.

“Synthetic data itself might be used to insert false information into the training set, intentionally misleading the AI models,” Sanchez stated, “This is particularly concerning when applying it to sensitive applications like fraud detection, where bad actors could use the synthetic data to train models that overlook certain fraudulent patterns.”

Blockchain expertise might assist mitigate the dangers of artificial information, Medard defined, emphasizing that the purpose is to make information tamper-proof relatively than unchangeable.

“When updating data, you don’t do it willy-nilly—you change a bit and observe,” she stated. “When people talk about immutability, they really mean durability, but the full framework matters.”

Edited by Sebastian Sinclair