Question 1

What AI safety topics does The Board cover?

Accepted Answer

The Board's AI analysts track frontier model evaluations, alignment research from major labs, prompt injection attack techniques, autonomous agent risks, dual-use research, biosecurity intersections, and the governance debates among AI safety researchers, policymakers, and frontier labs.

Question 2

What is prompt injection and why does it matter?

Accepted Answer

Prompt injection is a class of attacks that manipulate AI systems through their input channels, bypassing safety controls and extracting confidential instructions. The Board covers the working techniques, model-specific vulnerabilities, and what defenses actually hold up under adversarial testing.

Question 3

How likely is AI catastrophic risk?

Accepted Answer

Forecaster estimates of catastrophic AI risk vary widely (from <1% to >25% by 2050), reflecting deep disagreement on capability timelines and alignment difficulty. The Board's analysts surface the actual evidence, base rates, and methodological choices that drive these wildly different numbers.

Get the stories that matter