In the sophisticated, unforgiving world of Silicon Valley, where chaos is a credo and youth is both currency and burden, Suthir Balaji stands out as someone who questions the very foundations of the empire he has built. Ta. He was just 26 years old and a researcher working at OpenAI, one of the most influential AI companies on the planet. Nevertheless, rather than riding the wave of AI euphoria, he chose to oppose it. He expressed concern that the systems he helped create, specifically ChatGPT, were fundamentally flawed, ethically questionable, and legally questionable.
His tragic death in December 2024 shocked the tech industry. But at the same time, many were forced to face the uncomfortable truths he had been raising all along.
Just a kid who dared to ask a giant question
Balaji was not your typical Silicon Valley visionary. He wasn’t a grizzled founder with a decade of battle scars, or a nagging tech buddy proclaiming himself the savior of humanity. He was just a kid, albeit a super smart one, who just graduated from UC Berkeley in 2020 and started working at OpenAI.
Like many others in his field, he was fascinated by the potential of artificial intelligence, the dream that neural networks could solve humanity’s biggest problems, from curing disease to tackling climate change. Ta. For Balaji, AI was more than just code, it was a kind of alchemy, a tool that turned imagination into reality.
But by 2024, that dream had solidified into something darker. What Balaji saw in OpenAI, and its most famous product, ChatGPT, was a machine that was exploiting humanity instead of helping it.
ChatGPT: Vandal or thief?
ChatGPT was, and still is, a marvel of modern technology. Create poetry, solve coding problems, and explain quantum physics in seconds. However, behind the charm lies a deep and controversial truth. ChatGPT, like all generative AI models, was built with tons of input data collected from the internet, including data that is protected by copyright.
Balaji’s criticism of ChatGPT was simple. It meant that we were too dependent on the efforts of others. He claimed that OpenAI trains its models using copyrighted material without permission, infringing the intellectual property rights of countless creators, from programmers to journalists.
The process to train ChatGPT works as follows.
Step 1: Enter your data – OpenAI collected a large amount of text from the Internet, including blogs, news articles, programming forums, and books. Some of this data was publicly available, but much of it was copyrighted.
Step 2: Train the model – AI has learned how to analyze this data and generate human-like text.
Step 3: Generate output – When you ask ChatGPT a question, it doesn’t spit out an exact copy of the text it was trained on, but its responses often draw heavily from patterns and information in the original data.
Herein lies the problem, as Balaji saw it. Although the AI may not directly copy the training data, it still relies on it in a way that puts it in competition with the original creator. For example, asking a programming question to ChatGPT is likely to generate answers similar to those found on Stack Overflow. result? People stop visiting Stack Overflow, and the creators who shared their expertise there lose traffic, influence, and income.
Lawsuits that could change AI forever
Balaji was not alone in his concerns. In late 2023, the New York Times filed a lawsuit against OpenAI and its partner Microsoft for illegally using millions of articles to train its models. The Times claimed that the misappropriation caused direct harm to its business.
Mimic content: ChatGPT produces summaries or paraphrases of Times articles that may effectively compete with the original articles.
Market impact: AI systems threaten to replace traditional journalism by producing content similar to that produced by news organizations.
The case also raised questions about the ethics of using copyrighted material to create tools that compete with the very sources they rely on. Microsoft and OpenAI defended their practices, arguing that their use of data falls under the doctrine of “fair use.” This argument hinges on the idea that the data has been “transformed” into something new, and that ChatGPT does not directly reproduce copyrighted works. However, critics, including Balaji, considered this justification to be flimsy at best.
What critics say about generative AI
Balaji’s criticism fits into a larger story of skepticism about large-scale language models (LLMs) like ChatGPT. The most common criticisms are:
Copyright infringement: AI models scrape copyrighted content without permission and violate the rights of creators.
Damage to the market: By offering AI-generated alternatives for free, these systems devalue the original work, such as a news article, programming tutorial, or creative writing.
False alarm: Generative AI often produces “illusions” (fabricated information presented as fact), undermining trust in AI-generated content.
Opacity: AI companies rarely disclose what data their models are trained on, making it difficult to assess the full scope of potential copyright infringement.
Impact on creativity: Because AI models imitate human creativity, original creators may be shut out and the internet may be flooded with regurgitated derivative content.
Balaji’s Vision: Demanding Accountability
What set Balaji apart was not just his critique of AI, but the clarity and persuasiveness of his arguments. He believed that the unchecked growth of generative AI posed an immediate, not a hypothetical, danger. As more people rely on AI tools like ChatGPT, the platforms and creators that power the internet’s knowledge economy have been sidelined.
Balaji also argued that the legal framework governing AI is hopelessly outdated. U.S. copyright law, which was enacted long before the rise of AI, does not adequately address issues around data scraping, fair use, and market harm. He called for new regulations that would allow AI innovation to flourish while ensuring creators are fairly compensated for their contributions.
A legacy of questions, not answers
Suthir Balaji was neither a technology giant nor a revolutionary visionary. He was just a young researcher struggling with the meaning of his research. He spoke out against OpenAI, forcing his peers, and the world, to confront the ethical dilemmas at the heart of generative AI. His death is a reminder that the pressures of innovation, ambition and responsibility can weigh on even the brightest minds. But his critique of AI lives on and raises fundamental questions. As we build smarter machines, are we being fair to the humans who make their existence possible?
