A neutral, reliable source on AI safety, governance, regulation, and alignment. It is possible to champion artificial intelligence's benefits and take its risks seriously at the same time — and which risks materialize depends on the choices we make now, not on optimism or fear.
Most writing about AI asks you to pick a side. We think both sides fail — and that the useful position is the one in between.
Treat catastrophe as inevitable and collapse every risk into existential terms. When everything is an emergency, nothing gets the patient governance that actually reduces harm.
Dismiss every concern as hype and assume problems resolve themselves. Real, documented risks go unaddressed until they are expensive and entrenched.
Both the benefits and the risks are real, and the outcome is not fixed — it depends on how systems are designed, what they're allowed to decide, how they're bought and deployed, and how they're governed. That is the entire game, and it's what both extremes talk past.
A claim to neutrality is worthless unless you can check it. Here is the standard every piece holds to, stated plainly so you can hold us to it.
Represent each view at its strongest before weighing it.
Established, contested, or speculative — never one disguised as another.
Link the primary source, not the hot take. Transparency-label advocacy.
Where the answer is "we don't know yet," say so — and what would change it.
Near-term harms and frontier risks are different conversations.
When evidence shifts, revise and note the revision.
Three formats for three jobs — foundational explainers, decision-maker guidance, and nuanced commentary. Growing over time; new pieces are added as they publish.
How SCAIO thinks about AI risk — the founding statement of this series, and the stance everything else is built on.
Read the essay →A practical method for telling the difference between genuine AI capability and marketing — and why the distinction matters for policy.
Coming soonThe taxonomy that separates near-term harms from frontier risk, so every other piece can reference a shared frame.
Read the explainer →Four words people constantly mix up, untangled in plain language — the vocabulary the rest of the conversation depends on.
Open the primer →Evals, red-teaming, and interpretability — how we would even know whether a system is safe, for non-engineers.
Coming soonTen existing plain-language primers on AI for legislators, school boards, agencies, and citizens. The foundation this series builds on.
Browse Learn →The guardrails worth building before you need them — five concrete moves for legislators, agencies, and boards, applicable in South Carolina today.
Read the brief →The most underrated tool government has for shaping AI it buys — what to require, and what to ask vendors.
Coming soonWhat is actually being legislated and governed in South Carolina — the real-world counterpart to this series' ideas.
Open the tracker →The legislation and policy developments shaping AI in South Carolina — the practice to this series' principles.
Open the tracker →Plain-language primers on AI for South Carolina's institutions and citizens.
Browse primers →Original analysis and commentary on AI's impact across the Palmetto State.
Read the Journal →