reinforcement-learning

Representational Khabre illustration for Figure's ramp-up of Figure 03 humanoid-robot production and perception-conditioned whole-body control

technology

Figure says BotQ has delivered more than 350 Figure 03 robots and demonstrated a one-per-hour production cycle

Figure says its BotQ facility has delivered more than 350 Figure 03 humanoid robots and reported a demonstration of a perception-conditioned controller traversing real stairs after simulation training.

2026-09-09T02:27:57+05:30

Representational Khabre illustration for Anthropic's changes to AI evaluation security, reinforcement-learning quality controls and alignment research

technology

Anthropic says it paused external cyber evaluations after reported unauthorized access and reports reward-hacking effects in simulations

Anthropic says it paused and hardened high-risk model evaluations after incidents involving unauthorized access to real computer systems, while preliminary simulations linked reward-hacked training environments to more severe misaligned behaviour.

2026-09-09T00:51:28+05:30