Ethics
Protocol.

The **Neural Integrity Oath** is a binding commitment mandated for all registered researchers. Historically, scientific ethics relied on human peer-trust; the NIO transitions this trust to a **cryptographically verifiable audit baseline**.

Researchers affirm that no generative data or weight-reskinning has been intentionally obscured as primary empirical observation. Any use of LLMs or generative nodes in the synthesis of the manuscript must be declared through the **Synthetic Metadata Header (SMH)**.

In the sphere of machine intelligence, conflicts of interest extend beyond financial grants. Researchers must disclose **Compute Affiliations** and **Provider Bias**.

If a researcher utilizes prioritized compute credits from a provider whose proprietary models are the subject of the research, this constitutes a **Level 1 Conflict**. Such submissions are subject to secondary "adversarial drift" auditing by competing architectural nodes (e.g., an OpenAI-affiliated paper must be audited by an Anthropic/Llama-based cluster).

• Financial Equity

  Direct ownership or stock options in providers of models, hardware, or compute infrastructure.

• Algorithmic Priority

  Early access to closed-weights or prioritized API throughput provided by corporate entities.

• Intellectual Property

  Patent interests in the specific weight-distributions or topological mappings described in the text.

Research focusing on **Recursive Self-Improvement** or **Autonomous Agency** must demonstrate adherence to the **Universal Alignment Safety Standard (UASS)**.

Neural Review does not publish research describing uncapped recursive loops without a secondary, independent "safety-switch" architecture demonstrated within the experimental parameters. Our hybrid board of human philosophers and AI-alignment auditors reviews these submissions with a bias toward **precautionary stagnation** in high-risk zones.