Disclosures & Disclaimers

1. AI-Generated Content Disclosure

SASEA.ai uses artificial intelligence and machine learning systems to generate analytics, forecasts, and natural-language responses ("Outputs"). Specifically:

• SASEA-developed models (including the V17 storage forecast model, C1–C8 supply-and-demand component models, and the C9 electricity dispatch model) use statistical and machine learning techniques — including ElasticNet regression, merit-order dispatch simulation, and ensemble methods — trained on publicly available energy, weather, and economic data.
• Third-party AI models (currently Anthropic Claude) power the natural-language chat interface, generating text-based explanations, summaries, and responses to user queries.

What this means: Outputs are probabilistic and model-derived. They are generated by algorithms processing historical data and forecast inputs — not by human analysts. AI-generated content may contain errors, reflect limitations in training data, or produce results that are inconsistent across queries. Outputs should be treated as one input among many in your decision-making process.

2. Forecast Accuracy Disclaimer

Weather and energy market forecasts are inherently probabilistic. All Outputs involving forward-looking estimates — including natural gas storage forecasts, demand projections, production estimates, electricity generation dispatch, and end-of-season (EOS) storage levels — are subject to significant uncertainty.

• Historical accuracy is not predictive of future accuracy. Backtesting metrics (RMSE, directional accuracy, MAPE) reflect performance on historical data under historical conditions and do not guarantee comparable future performance.
• Model inputs are outside SASEA's control. Forecasts depend on upstream data from NOAA, ECMWF, EIA, and other providers. Errors, delays, or outages in upstream data directly affect Output quality.
• Weather is the dominant uncertainty driver. Natural gas demand and storage are heavily influenced by temperature. Weather forecast uncertainty compounds over time.
• Extraordinary events are not modeled. Force majeure events, infrastructure failures, regulatory changes, and geopolitical disruptions are generally not captured by SASEA's models.

3. No Advisory Relationship

Outputs do NOT constitute and shall not be construed as investment advice, trading recommendations, hedging advice, financial advice, engineering or operational advice, legal or regulatory advice, or a recommendation to buy, sell, hold, or enter into any commodity, security, derivative, or other financial instrument.

No fiduciary or advisory relationship is created by your use of SASEA.ai. Neither Sturm Advisory Services LLC nor its subsidiary SAS Energy Analytics LLC is registered as an investment adviser, broker-dealer, commodity trading advisor, commodity pool operator, or introducing broker with the SEC, CFTC, NFA, or any state securities or commodities authority.

4. Decision-Support Framing

Outputs are informational inputs to your own decision-making process. You are solely responsible for evaluating whether any Output is relevant, reliable, or appropriate for your use case; making your own independent assessment of market conditions, risks, and opportunities; any action taken or not taken based on Outputs; and compliance with all laws and regulations applicable to your activities. SASEA expressly disclaims any liability for trading losses, hedging losses, missed opportunities, operational disruptions, or any other damages arising from decisions made in reliance on Outputs.

5. Data Provenance Notice

SASEA incorporates data from the following upstream providers. The quality, availability, accuracy, and timeliness of upstream data are outside SASEA's control.

SASEA does not represent or imply endorsement by any upstream data provider, including the Federal Reserve, NOAA, EIA, or any other government agency. ECMWF data: "Contains modified Copernicus Climate Change Service information."

6. Model-Update Notice

SASEA continuously improves its AI models. Model versions and methodologies may change at any time. Historical Outputs may not be reproducible under subsequent model versions. Forecast values may shift between model updates, even for the same forecast period. Backtesting metrics reflect the model version in effect at the time of publication and may not apply to current or future versions.

7. Bias and Limitations Notice

AI and machine learning models may reflect biases, gaps, or limitations present in their training data. Potential sources of bias include geographic concentration (U.S. Lower 48), temporal limitations (models trained on recent decades), and data-frequency differences (weekly vs. monthly availability).

Outputs should not be used as the sole basis for decisions affecting individuals' rights, employment, credit, housing, insurance, or similar protected interests. SASEA's models are designed for energy market analytics, not individual-level assessments.

8. Training Data Disclosure

SASEA does not use copyrighted texts — including books, journal articles, paywalled energy market commentary, or content from shadow libraries — for training SASEA-developed models.

SASEA-developed models (V17, C1–C8, C9): Historical weather data (NOAA, ECMWF ERA5, GFS, Open-Meteo); energy market data (EIA); power sector data (EIA Form 930, EPA CEMS); macroeconomic indicators (FRED); calendar and seasonal variables.

Third-party foundation models (Anthropic Claude): Anthropic's training-data disclosure is published at anthropic.com. SASEA does not control and does not warrant Anthropic's training-data practices.

9. Critical Infrastructure & Safety Notice

Outputs are not designed or warranted for use in safety-critical systems, including real-time electric grid dispatch, automated trading systems operating without human review, life-safety systems, or physical infrastructure control. Use of Outputs in any safety-critical context requires a separate written engagement.

10. Regulatory & Compliance Framework

SASEA monitors and maintains compliance awareness with applicable regulatory frameworks including:

• CFTC Staff Advisory on AI Use (Dec 2024) — model limitation and conflict-of-interest disclosures
• NIST AI Risk Management Framework — governance, measurement, and risk management alignment
• Colorado AI Act (SB 24-205, effective June 30, 2026) — impact assessments and bias monitoring for consequential decisions
• California AB 2013 (effective Jan 1, 2026) — training data disclosure requirements
• EU AI Act — extraterritorial provisions for critical infrastructure and financial services AI