Step inside a documented incident at the moment a decision matters, make the call, see how you scored, and earn a certificate anyone can verify. No signup, no cost, ever.
Every module puts you inside a real incident at the decision point. All 18 are free.
Read the real cases: the $25M deepfake CFO - Samsung's ChatGPT leak - the fake case-law filing.
AI screening discriminates - and liability can reach employer and vendor.
Play free →One email quietly tells Copilot to hand over your data. Zero clicks (EchoLeak, CVE-2025-32711).
Play free →Poison the training data, poison the model. Tay learned it in 16 hours.
Play free →MFA fatigue: an hour of pushes, a fake IT message, one tap. Uber, 2022.
Play free →A vendor's new bank details, a queued wire, 20 seconds to decide. BEC cost $2.77B in reported US losses, FBI 2024.
Play free →


Training evidence designed with SEC Reg S-P, NYDFS 500.14, and FTC Safeguards in view - the duty stays with your firm.
Enter your firm name and we'll generate an AI Use, Training & Incident-Readiness Policy with your firm's name already in it - ready to add to your policy manual today. It documents a training plus incident readiness program with scored completion, verifiable certificates, and firm-level reporting, aligned with the training expectations under Reg S-P, NYDFS 500.14, and FTC Safeguards.
All 18 drills are free, and the Monday Brief is out. Join the list for the weekly Brief and product drops.
Pick a plan, send one link, export the evidence.
All 18 incident drills, free for every learner.
Your team trains on any phone or laptop.
Scores and decision paths per employee, ready for the exam. LMS-ready exports for the training system you already run.
What happened, what it means for your policy, in a 3-minute read. Same watch that feeds the course.
Per-employee completion, timestamps, certificate IDs, team-level results - plus a team certificate when everyone finishes. This is the evidence export every customer gets.
Cyber and E&O applications are starting to ask about AI training. A policy document isn't a training record.
Our reading of existing books-and-records and supervision rules: AI-generated notes, summaries, and client comms fall under them. Training shows the supervision was real.
If your firm wrote an AI policy in the last 18 months, did anyone attach training to it? Close that gap this week.
The annual deck gets clicked through and forgotten within weeks. Ten minutes inside a real incident builds the reflex instead. The learning science behind every choice →
Learners make the call inside the incident.
Short drills repeat the safe move under time pressure.
Skepticism and control, tracked per learner.
Completion, scores, timestamps, and certificate IDs export per employee.
New modules monthly; the Brief covers one every Monday.
One incident, one skill, one certificate in 10-25 minutes.
Marketing, finance, engineering, HR, and leadership.
Sid Yenamandra founded Entreda (acquired by Smarsh) and spent a decade on the vendor side of SEC and FINRA exams. He is also Founding Managing Partner of Surge Ventures and Founder & CEO of SurgeONE.ai. AI Safety 101 turns real incidents into short drills with verifiable evidence for the training file. Berkeley EE/CS. Silicon Valley engineer first. More about the founder →
AIS101 Firm: role-based curriculum built from your team's real drill data, the firm portal to assign and track, and verifiable evidence for your compliance file. Individual answers and named low scores are never exposed.
Scoped with your team, never off a rate card.
Half-day virtual session for up to 15 leaders.
90 minutes on what AI can touch, what it must never touch, and who owns it.
Prompt injection, data exfiltration, and agent misuse, delivered with a specialist partner.
Your policies and incident types as a scored drill with the same evidence file.
Not yet. Data is hosted on Supabase. We store learner name, email, company code, scores, decision paths, and timestamps. We do not sell data or use learner responses to train third-party models. DPA available.
Yes. Sources are shown inside each module.
No vendor can satisfy it for you. The course produces completion records, scores, and verifiable certificates your firm can retain as training evidence.
No. Test first, then teach. Learners make the call inside a real incident, get scored, then learn in the debrief.
Short training that teaches people what data never goes into AI, how to check AI output, and what to do when something looks wrong.
Broad platforms cover general security awareness. AI Safety 101 goes deep on AI incidents with scored decision drills and evidence. Many firms run both. Competitor scope as of September 2026.
No. All 18 drills are free for anyone. Regulated firms get the most from the evidence layer.
New modules every month from documented incidents, plus the Monday Brief.
Make sure your team trained on the last one - with training they'll actually finish.
We come back within one business day with times to talk.