Apple and politics: New allegations surrounding AI guidelines

Apple is facing a new controversy. A leak about its artificial intelligence training guidelines is sparking speculation that Apple changed its guidelines to accommodate President Trump. These allegations are based on a Politico report comparing the guidelines updated in March 2025 with the 2024 version. While Apple denies any political motivation, the report demonstrates how quickly technical processes can be placed in a political context.

Artificial intelligence only works reliably when humans train the models and evaluate outputs. Apple works with subcontractors like Transperfect, which employs around 200 people in Barcelona. These people review input requests, monitor outputs, and apply Apple's guidelines. A leak of these documents has sparked a public debate that says more about political interpretation than about the actual work being done on Apple's AI systems.

Politically motivated changes?

Politico argues that Apple adjusted its policies in March 2025 to accommodate the Trump administration. The report highlights that terms like "systemic racism" were removed and replaced with more general language about discrimination. Topics such as diversity, equity, and inclusion are marked as controversial. Vaccines and elections are also new to the list of sensitive topics. The political term "radical" is given greater weight and classified as inflammatory. In addition, discussions about Gaza must be treated separately. Particularly striking is the increased number of mentions of Trump: from three in the old document to eleven in the new version. Politico interprets this as a signal of a deliberate adjustment. However, the context of these mentions remains unclear.

Apple's response to the allegations

Apple strongly denies the allegation. In a statement to Politico, the company emphasizes that Apple Intelligence is based on clear principles for responsible AI. These principles guide every step, from development to training and evaluation of the models. Changes to the guidelines are common and serve the purpose of continuous improvement. Subcontractor Transperfect also clarifies that guidelines can be adjusted up to 70 times a year. Neither Apple nor Transperfect see this as a fundamental change in policy, but rather a normal process for updating sensitive topics.

What the changes really mean

The leaked guidelines demonstrate one thing above all: an adaptation to current debates. Controversial topics will be monitored more closely to ensure that the AI responds responsibly. For contractors, this means working with greater care, context, and evidence when sensitive topics arise. Furthermore, content based on copyrighted material or disparaging Apple executives will be flagged. Both are industry standard and neither specific nor controversial. Rather, they seem like a precautionary measure to avoid legal problems or targeted attacks.

Subcontractors in focus

The discussion surrounding Apple's guidelines highlights how deeply subcontractors are involved in the development process. Employees are often said to be unaware of who exactly they are working for. Nevertheless, there is sufficient evidence that their evaluations flow directly into Apple's systems. The Politico report suggests that this could be part of the training for a new chatbot. However, Apple executives like Greg Jozwiak clarify that no dedicated chatbot is planned. Instead, it could be further developments of Siri, possibly in conjunction with services like Gemini-based web agents.

Between media criticism and reality

Politico frames its criticism in such a way that Apple's actions seem more dramatic than they actually are. The lengthy report contains many tangents that obscure the core points. In the end, it seems as if the focus is less on concrete changes and more on generating outrage over Apple's perceived capitulation to Trump. It's clear, however, that Apple hasn't changed course. The company continues to invest heavily in diversity, green energy projects, and international trade—issues that are significantly different from Trump's agenda.

Apple's line: responsibility instead of political agenda

The debate over Apple's AI guidelines demonstrates how closely intertwined technology and politics have become. A normal adjustment process, as is common in AI development, can easily be interpreted as a political signal. Apple itself remains committed to its policy: continuous improvements, responsible guidelines, and clear principles for dealing with sensitive issues. Accusations of a Trump-friendly approach are based more on speculation and media exaggeration than on hard facts. (Image: Shutterstock / RYO Alexandre)