Fact-Checking & Source Verification Policy
Effective date: September 2026 • Editorial Desk
1. Primary Source Mandate
At AI Launch Watch, we enforce a strict primary source verification mandate. We do not rephrase unverified blog posts or secondary press releases. Every technical assertion, benchmark score, or pricing claim must trace back to one of the following primary sources:
- Official Developer Documentation: Public API specs, SDK repositories, and developer release changelogs.
- Published Research Papers: arXiv preprints, peer-reviewed AI journal papers, and model technical report whitepapers.
- Direct Checkout Portals: Public pricing pages, enterprise SLA contracts, and billing documentation.
- Verifiable Code Repositories: Open-source GitHub/GitLab repositories with commit histories and issue tracker records.
2. Zero Tolerance for Fake Testing & Marketing Fluff
Artificial intelligence marketing frequently uses hyperbolic buzzwords like "human-level", "game-changing", or "revolutionary". Our editorial guidelines explicitly forbid accepting marketing buzzwords as empirical facts:
When a vendor publishes benchmark scores (such as MMLU, HumanEval, or MATH), we explicitly indicate whether the results were evaluated via 0-shot, 5-shot, or chain-of-thought prompting. If results are vendor-reported rather than independently reproduced, we label them as "Vendor Reported".
3. Audit Cadence & Price Verification
Because cloud inference pricing and token models fluctuate frequently, our automated content auditor continually tracks product pages:
- Top trending tools undergo bi-weekly pricing and model parameter checks.
- Directory listings include explicit timestamp dates (
Last Price Check,Specs Verified). - Deals, coupon codes, and limited-time promos are audited weekly for expiration.