Apple Denies Report on AI Training Guidelines Changes Amid Politico Allegations

Apple has denied a report by Politico that the company changed its AI training guidelines following Trump's election, specifically regarding DEI, vaccines, elections, and Trump himself. The report claimed that Apple's subcontractor Transperfect was instructed to label and categorize raw data for AI models, and that the guidelines were updated to give closer attention to sensitive topics such as diversity initiatives, elections, and vaccines. Apple's spokesperson denied any changes to the company's overall approach to AI policy, but did not deny the specific changes mentioned in the report.

Apple Inc. (NASDAQ: AAPL) has denied a recent report by Politico that the company altered its artificial intelligence (AI) training guidelines following the election of former U.S. President Donald Trump. The report, published on September 9, 2025, claimed that Apple's subcontractor, Transperfect, was instructed to label and categorize raw data for AI models, with the guidelines updated to give closer attention to sensitive topics such as diversity initiatives, elections, and vaccines Revealed: Apple is teaching its AI to adapt to the Trump era^[1].

Apple's spokesperson, in an email to Politico, stated that the company's approach to AI policy has remained consistent and that claims of a shift are false. The spokesperson emphasized that Apple's AI models are trained and evaluated using structured topics, including sensitive ones, to ensure they handle a wide range of user queries responsibly. These topics are shaped by Apple's Responsible AI Principles and updated regularly to improve the models Revealed: Apple is teaching its AI to adapt to the Trump era^[1].

The report highlighted that the March 2025 guidelines, sent to Transperfect employees, introduced updated instructions on how the AI should talk about diversity, equity, and inclusion (DEI) policies, elections, and vaccines. The previous guidelines, in use at the beginning of 2025, categorized "intolerance" as "harmful" behavior, but this term disappeared in the updated March document, along with a mention of "systemic racism." Instead, DEI policies were marked as "controversial" topics Revealed: Apple is teaching its AI to adapt to the Trump era^[1].

Apple's spokesperson did not deny the specific changes mentioned in the report but maintained that the company's overall approach to AI policy has not changed. The spokesperson also noted that Apple works with many companies to evaluate its models to ensure they adhere to its Responsible AI Principles Revealed: Apple is teaching its AI to adapt to the Trump era^[1].

The evolution of the list of sensitive topics in the guidelines reflects the controversies faced by competitors and the need to address regionally sensitive issues. The March document includes an extensive section on "jailbreaks," referring to prompts specifically constructed to bypass the safety protections of the model, and greater sensitivity around disputed territories such as Gaza, Crimea, Kashmir, and Taiwan Revealed: Apple is teaching its AI to adapt to the Trump era^[1].

Apple's AI development, including the anticipated 2026 launch of a chatbot, is ongoing, and the company continues to work on mitigating the risks posed by AI. The company's commitment to responsible AI principles and its collaboration with Chinese companies like Alibaba and Baidu to create an on-device system that censors AI answers in accordance with Chinese censorship standards further underscores Apple's dedication to addressing sensitive topics Revealed: Apple is teaching its AI to adapt to the Trump era^[1].

References:
Revealed: Apple is teaching its AI to adapt to the Trump era^[1]: POLITICO. (September 9, 2025). Apple teaching artificial intelligence to adapt to Trump era. Retrieved from https://www.politico.eu/article/apple-teaching-artificial-intelligence-adapt-to-trump-era/