×
Apple Intelligence Beta Testers Find “Do Not Hallucinate” in System Prompts
Written by
Published on
Join our daily newsletter for breaking news, product launches and deals, research breakdowns, and other industry-leading AI coverage
Join Now

Apple’s forthcoming AI features, collectively known as Apple Intelligence, are undergoing beta testing, revealing the company’s approach to addressing common AI pitfalls and ensuring responsible implementation.

Uncovering Apple’s AI guidelines: Testers of the macOS Sequoia beta have discovered plaintext JSON files containing prompts designed to guide the behavior of Apple’s AI features.

  • These files are located in a specific folder on Macs running the beta version with Apple Intelligence enabled.
  • The prompts provide valuable insights into Apple’s strategy for keeping its AI narrowly focused and factual.
  • Many of the instructions are utilitarian, describing the intended behavior for features like Smart Reply in Mail.

Combating AI hallucinations: Apple has implemented specific prompts aimed at preventing AI confabulations or hallucinations, demonstrating a commitment to accuracy and reliability.

  • Explicit instructions such as “Do not hallucinate” and “Do not make up factual information” are included in the prompts.
  • These directives highlight Apple’s efforts to maintain the integrity of information provided by its AI systems.
  • The approach aligns with broader industry concerns about AI-generated misinformation and the need for responsible AI development.

Feature rollout and hardware requirements: Apple Intelligence features are set for a phased release, with specific hardware requirements to ensure optimal performance.

  • The AI features will launch in beta this fall but will not be included in the initial iOS 18.0 and macOS 15.0 releases.
  • Users will need recent hardware to access these features, including iPhone 15 Pro or Macs and iPads with at least an M1 chip.
  • This hardware requirement suggests that Apple’s AI implementations may be computationally intensive, necessitating more powerful processors.

Balancing functionality and responsibility: Apple’s approach to AI development appears to prioritize both innovative features and responsible implementation.

  • The discovery of these prompts indicates Apple’s commitment to transparency in its AI development process.
  • By setting clear guidelines for its AI, Apple aims to deliver useful features while mitigating potential risks associated with AI-generated content.
  • This strategy may help Apple differentiate itself in the competitive AI landscape by emphasizing reliability and user trust.

Implications for the AI industry: Apple’s cautious approach to AI implementation could set a precedent for other tech companies and influence industry standards.

  • The explicit instructions against hallucination reflect growing concerns about AI accuracy and the potential for misinformation.
  • Apple’s strategy of running AI features on-device aligns with increasing focus on privacy and data protection in AI applications.
  • The company’s methodical rollout may pressure competitors to prioritize responsible AI development over rapid feature deployment.

Looking ahead: Potential impact and adoption: As Apple prepares to introduce these AI features, the tech industry and consumers alike will be watching closely to assess their effectiveness and reception.

  • The success of Apple’s approach could influence how other companies develop and implement AI technologies in consumer products.
  • User experiences with these features may shape public perception of AI capabilities and limitations in everyday applications.
  • Apple’s focus on preventing hallucinations and ensuring factual accuracy could raise the bar for AI reliability across the industry.
“Do not hallucinate”: Testers find prompts meant to keep Apple Intelligence on the rails

Recent News

Nvidia’s new AI agents can search and summarize huge quantities of visual data

NVIDIA's new AI Blueprint combines computer vision and generative AI to enable efficient analysis of video and image content, with potential applications across industries and smart city initiatives.

How Boulder schools balance AI innovation with student data protection

Colorado school districts embrace AI in classrooms, focusing on ethical use and data privacy while preparing students for a tech-driven future.

Microsoft Copilot Vision nears launch — here’s what we know right now

Microsoft's new AI feature can analyze on-screen content, offering contextual assistance without the need for additional searches or explanations.