WhatsApp's help pages contained hidden instructions related to large language models (LLMs), which were not intended for public view. These instructions appear to guide the AI's behavior but do not recommend the involvement of a regulator. The discovery raises questions about transparency and the internal workings of AI systems used by major platforms. AI
IMPACT Raises questions about transparency and control over AI behavior in consumer products.
RANK_REASON Discovery of hidden LLM instructions within a widely used product.
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →