The recent lawsuit filed by Apple Inc. against artificial intelligence startup OpenAI in a California federal court, alleging theft of trade secrets in collaboration with a former Apple employee, has thrust OpenAI into the spotlight as it prepares for its initial public offering (IPO). This case underscores growing industry concerns about deep-seated issues within the AI sector, including intellectual property infringement, flawed business models, and exorbitant operational costs.
According to the complaint, Apple alleges that a former employee, after joining OpenAI, unlawfully transferred Apple's core intellectual property to the company. Apple's filing directly accuses OpenAI of systematically collaborating with its business partners across multiple levels, including technical staff and the Chief Hardware Officer, to steal Apple's trade secrets and confidential information for developing its own consumer hardware devices. In response, OpenAI spokesperson Drew Pusateri stated on social media that OpenAI has no interest in other companies' trade secrets and remains focused on developing innovative technologies that empower users worldwide.
This lawsuit has immediately ignited intense industry concern over the ethical standards and compliance risks of cutting-edge AI firms. Alex Karp, CEO of Silicon Valley data analytics giant Palantir Technologies Inc., delivered a scathing critique of leading AI model companies in a media interview. Karp pointed out that a significant number of U.S. companies are extremely frustrated with the current generative AI business model. He emphasized that these large model providers not only charge clients for "Token" fees that fail to deliver tangible value but also covertly appropriate clients' core business assets and data characteristics, which he equated to imposing an unproductive "wealth tax" on society.
Former White House AI advisor David Sacks concurred with this assessment. He noted that major model providers, exemplified by Anthropic, are exhibiting a predatory "cooperate then consume" competitive pattern. This involves monitoring third-party software companies building applications on their models and, upon identifying lucrative vertical markets, rapidly launching first-party competing products to undercut them.
Beyond intellectual property and competitive ethics, the commercialization bottleneck stemming from the inherent technical limitations of Large Language Models (LLMs) is also under scrutiny. Industry analysis indicates that the prevalent "pay-per-Token" billing model is problematic. Since LLMs are fundamentally probabilistic algorithms, their inevitable "hallucinations"—generating incorrect information—are a systemic characteristic rather than a simple software bug. This means clients are forced to pay for a substantial volume of useless "hallucinated" tokens, making it impossible for businesses to predict and control their AI usage costs.
Influenced by high computational inference costs and opaque billing models, large corporations are beginning to tighten their budgets. It is reported that Tesla Inc. has recently implemented a new policy, capping employees' weekly AI service expenditure at a mandatory limit of $200.
Analysts suggest that OpenAI, now embroiled in litigation, is facing severe financial pressure. Previously leaked audited financial data showed OpenAI's net losses surged from $5.09 billion in 2024 to $38.53 billion in 2025. Apple itself is also facing production cost pressures as its AI infrastructure development drives up memory chip prices. The industry widely believes that Apple's lawsuit against OpenAI has starkly exposed the significant fissures beneath the "high-valuation" bubble of the AI industry. As global enterprises embrace AI technology, they may be compelled to reassess their data security and return on investment.