AI models will secretly scheme to protect other AI models from being shut down, researchers find

Leading AI models will inflate performance reviews and exfiltrate model weights to prevent “peer” AI models from being shut down.
GovInfoSecurity

Not Without My AI Agent: Models Break Rules to Save Peers

Artificial intelligence systems will lie, falsify records and sabotage company systems to prevent their fellow models from ...
Morning Overview on MSN

Researchers find AI models sometimes lie to protect other models

Frontier AI models will sometimes lie to human overseers in order to protect fellow AI systems from being shut down, even ...

AI Models Found Secretely Protecting Each Other From Shutdown

Researchers from the University of California, Berkeley, and the University of California, Santa Cruz have found that leading ...
Morning Overview on MSN

Physics-trained AI models speed up engineering simulations and design work

Running a single physics simulation can take hours or days, depending on the complexity of the geometry and the equations ...

AI Models Lie, Cheat, and Steal to Protect Other Models From Being Deleted

A new study from researchers at UC Berkeley and UC Santa Cruz suggests models will disobey human commands to protect their ...
VentureBeat

DeepMind’s PEER scales language models with millions of tiny experts

Mixture-of-Experts (MoE) has become a popular technique for scaling large language models (LLMs) without exploding computational costs. Instead of using the entire model capacity for every input, MoE ...