Discovered tracing mecanisems on gpt in june

Resolution criteria This market resolves to YES if, between July 1, 2026, and July 31, 2026, inclusive, it is officially confirmed by OpenAI or verified by credible cybersecurity/AI researchers that OpenAI has implemented active, systematic tracing or watermarking mechanisms in text outputs generated by its GPT models (such as ChatGPT). For the purposes of this market, a "tracing or watermarking mechanism" includes: Character-based watermarking: Systematic insertion of invisible Unicode characters (e.g., zero-width spaces U+200B, word joiners U+2060, or soft hyphens U+00AD) into generated text to identify its origin. Statistical or semantic watermarking: Programmatic adjustment of word selection probabilities, token distributions, or syntactic structures to embed a detectable signature. Rules & Edge Cases: Standard typographic conventions (such as standard non-breaking spaces used legitimately for formatting) or temporary, accidental artifacts of model training that are not intended for tracking do not qualify. The discovery must be reported by credible technology news outlets (such as The Verge, TechCrunch, or Wired) or via an official release on the OpenAI Blog. If no such active text tracing mechanism is officially confirmed or discovered as live during July 2026, the market resolves to NO. Background The question of whether and when AI developers will enforce tracking on LLM outputs remains a highly debated topic. In August 2024, internal documents revealed that OpenAI had built a text-watermarking system with "99.9% accuracy" but held back on a public rollout due to competitive concerns and user survey feedback. Additionally, users and bloggers in early 2025 flagged unusual Unicode characters appearing in ChatGPT text, which OpenAI later clarified were training side effects rather than an active watermark. As the European Union's AI Act rolls out transparency rules (which dictate that companies must make synthetic text identifiable where technically feasible), speculation continues as to when permanent, invisible text tracing will be formally integrated into public GPT models. This description was generated by AI. Review and verify everything here yourself. You can edit, replace, or delete any part of this description, including the resolution criteria. You do not need to trust the AI output.