Michigan's 'Explosive Diarrhea' Parasite Outbreak Passes 300 Cases, Six Times the Normal Rate

I asked an AI chatbot last month to help me calculate how many gallons of water my household needs for two weeks, accounting for drinking, cooking, and basic hygiene. It gave me a detailed breakdown in about 30 seconds that would have taken me 20 minutes to research manually. Then I asked it about a specific water filter and it gave me information about a model that doesn't exist. AI tools are incredibly useful. They're also confidently wrong sometimes. Knowing that they can be useful, and occasionally wrong is something to keep in the back of your mind.

Here's what I've got this morning.

AI as a Preparedness Research Assistant

AI chatbots (Grok, ChatGPT, Claude) are powerful research tools that can help with preparedness planning, calculations, scenario analysis, and information gathering. They're fast, available 24/7, and can process complex questions that would take hours to research manually. They also make mistakes, sometimes convincingly. Using them effectively requires understanding both their strengths and limitations.

What AI does well for preparedness: calculations (water needs per person per day times household size times duration), scenario planning ("what should I prioritize if I have $200 for preparedness supplies?"), organizing information (turning a list of supplies into a categorized, prioritized checklist), explaining concepts (how does a pressure canner work, what's the difference between FRS and GMRS), and drafting plans (family communication SOPs, evacuation checklists).

What AI gets wrong: specific product recommendations (it may describe products that don't exist or confuse model numbers), current events (its training data has a cutoff), local regulations (rainwater harvesting laws, GMRS licensing specifics), and medical dosages (never rely on AI for medication guidance without verifying with a professional source). Always verify specific claims, especially anything safety-critical.

The verification habit is essential. For any factual claim from an AI, ask: can I verify this with a primary source? USDA guidelines, manufacturer websites, government publications, and established preparedness resources (ThePrepared.com, Extension.org) confirm or correct AI outputs. Use AI to generate the first draft. Use primary sources to verify.

AI excels at personalization. "I live in a third-floor apartment in Houston with two adults and a cat. What should my water storage plan look like?" produces tailored advice that generic guides don't offer. The more context you provide, the more specific and useful the output.

For ongoing preparedness, AI can help you interpret data (explaining what the NY Fed Supply Chain Index means), analyze scenarios (what happens if a Category 3 hurricane hits your region), and create custom checklists (seasonal kit swap lists for your specific climate).

Ask an AI chatbot to create a personalized preparedness checklist for your specific household.

Provide details: household size, ages, medical needs, living situation, climate, and budget. Ask for a prioritized checklist. Then verify the top 5 recommendations against a trusted source. You'll get a useful starting point and learn the verification habit.

AI-Generated Phishing Surged 14x in a Single Month — Now More Than Half of All Detected Attacks

Hoxhunt's 2026 Phishing Trends Report, drawn from millions of reported emails across 2.5 million users in 131+ countries, found AI-generated phishing jumped from under 5% to 56% of all detected attacks in a single month at the end of 2025. Separately, malicious packages planted in public code repositories climbed from 55,000 in 2022 to 454,600 in 2025, and the average time from a vulnerability's public disclosure to a working exploit in the wild has collapsed from over 700 days in 2020 to just 44 — with more than a quarter of vulnerabilities now exploited within 24 hours of disclosure.

Preparedness angle: the old "check for typos" rule is dead. Verify any unexpected request for money, login credentials, or personal information by calling the sender directly, using a number you already have on file, never one provided in the email itself.

Source: The Hacker News, Hoxhunt

Jessie Krebs' SERE instruction career was built on the principle that tools are only as good as the person using them. On Hacking the Wild, she demonstrated that any tool, from a knife to a fire starter, requires understanding its capabilities and limitations to be effective.

AI tools follow the same principle. They're powerful when used within their strengths (research, calculation, organization, explanation) and dangerous when trusted beyond their capabilities (specific product details, medical advice, current events). The prepared user knows what to ask AI for and what to verify elsewhere.

Michigan's Parasite Outbreak Triggers an 'Explosive Diarrhea' Warning — Cases Climb Past 300, Six Times the Normal Rate

Michigan health officials have confirmed more than 300 cases of cyclosporiasis this year in a state that typically sees about 50 annually. The CDC has separately tracked another 145 cases across 17 other states, with at least 20 hospitalizations nationwide. Cyclospora is a microscopic parasite usually spread through contaminated fresh produce — berries, herbs, and leafy greens are common culprits — and it causes watery, explosive diarrhea that can persist for weeks if untreated.

Preparedness angle: rinse fresh produce thoroughly, and keep an anti-diarrheal and oral rehydration salts in your medical kit. Untreated cyclosporiasis can cause serious dehydration, especially in children and the elderly.

Source: ABC7 Chicago, NBC News, CDC

AI-Generated Emergency Plan Review

I pasted my family emergency plan into an AI chatbot and asked it to identify gaps. It flagged three things I'd missed: no plan for pet care during evacuation, no backup meeting point if the primary was inaccessible, and no protocol for reuniting if family members were separated in different vehicles.

All three were legitimate gaps. The AI spotted them in about 10 seconds because it compared my plan against comprehensive emergency planning frameworks. I then verified its suggestions against Ready.gov and FEMA templates. All three recommendations were sound. Free.

Budget alternative: Use the free version of any AI chatbot. The preparedness planning capabilities are available on free tiers. Just bear in mind that free models are usually the weaker, older versions that aren’t as smart as the newer ones.

Overrated: AI-generated product recommendations. AI chatbots frequently recommend products with incorrect model numbers, discontinued items, or features that don't exist. Always verify product recommendations on the manufacturer's website or a trusted review site before purchasing.

Underrated: AI for interpreting data you've collected. Upload your pantry inventory spreadsheet and ask "how many days can my household of four sustain on this food supply?" The calculation is instant and more thorough than manual counting.

Claude.ai — Anthropic's AI assistant. Useful for research and planning.

Ready.gov — Federal preparedness verification source.

Extension.org — University-verified guidance on food, water, and home topics.

Grokipedia — Reference source for background verification.

Improvisation: making do with what you have. The mindset and techniques for solving problems when you don't have the right tools.

PS: I use AI as the first step in most preparedness research, then verify everything that matters against primary sources. The combination is faster and more thorough than either approach alone. The key is never skipping the verification step.