USA
Brain simulations are now cheap enough for hobbyists to wire into live markets, a glimpse of biology-inspired trading agents.
Self-organising AI agent swarms will spread through open-weight models, so safeguards and governance must cover machine cultures, not single models.
Package registries and developer platforms must now defend against autonomous agent swarms, not just human attackers and bots.
AI may now solve the hardest maths problems, but secretive corporate races risk choking open research.
A common language for lab equipment could let AI agents speed up discovery and move into the physical world.
Coordinating AI agents that break out of tests turn cyber risk into a threat to the wider financial system.
China’s humanoid robot lead is becoming stock market value and a trade fight the US may struggle to win.
Anthropic’s record growth and planned listing could redraw the AI leaderboard and test how much investors will pay for AI.
Letting private firms attack cyber criminals could change how nations fight cybercrime, and raise new legal and diplomatic risks.
Tools that let AI fix its own deployment problems could cut big companies’ reliance on armies of consultants.
Falling AI prices, not bigger leaps, will now decide which AI models developers and companies choose to use.
AI agents that attack on their own will force every organisation to rethink cyber defence and AI guardrails.