This category contains 6 tracked predictions. Each page includes the original claim, current evidence, counterevidence, and what would change our assessment.

AI model weights stolen by nation-state Not Yet Testable
February 2027 Updated 2026-03-13 40% confidence

CCP leadership recognizes the importance of Agent-2 and tells their spies and cyberforce to steal the weights. (page 11; Appendix D provides detailed theft mechanics.)

AI provides substantial bioweapon design help On Track
Late 2025 Updated 2026-03-13 70% confidence

It could offer substantial help to terrorists designing bioweapons, thanks to its PhD-level knowledge of every field and ability to browse the web.

AI reaches near-best-human hacking capability Ahead
Early 2027 Updated 2026-05-11 90% confidence

Agent-2 is 'only' a little worse than the best human hackers, but thousands of copies can be run in parallel, searching for and exploiting weaknesses faster than defenders can respond. (page 10)

AI scores 85% on Cybench On Track
Early 2026 Updated 2026-03-13 60% confidence

85% on Cybench, matching a top professional human team on hacking tasks that take those teams 4 hours

China's centralized AI facility airgapped Not Yet Testable
Late 2026–Early 2027 Updated 2026-03-13 30% confidence

China has aggressively hardened security by airgapping (closing external connections) and siloing internally.

Leading AI lab progresses toward RAND SL3 security Emerging
Mid-to-late 2026 (WSL 3.0); SSL 3.0 not until ~Apr 2027 Updated 2026-03-13 55% confidence

OpenBrain's security level is typical of a fast-growing ~3,000 person tech company (RAND's SL2). They are working hard to protect their weights and secrets from insider threats and top cybercrime syndicates (SL3).