ML News

Attackers can bury a second network inside a model's weights, encrypted so it runs only when a secret trigger appears. Fine-tuning, weight noise, and trigger searches failed to surface it, and mathematically it stays hidden even to anyone reading the full weights.

Instead of retraining the AI's brain, this framework lets an agent rewrite its own prompts, memory, tools, and code, keeping the changes for later runs. The analysis shows such lasting edits stay reliable only if each change passes fresh, independent checks first.

A platform that gives AI agents many isolated, stateful environments for training, mixing lightweight and heavy isolation with shared, reusable parts loaded on demand. One cluster unit runs about 3 million sandboxes a day, cutting setup and image-loading costs versus a single sandbox runtime.

Training a hybrid AI to reason in its no-thinking mode seems to sharpen its skills, but the gains come from switching its thinking mode back on. Swapping out one internal signal erased most of the improvement, and accuracy never beat the thinking mode.

Instead of outside code deciding what stays in memory, the model treats its context as a file it can freely edit. On a hard web-search task it beat the best outside strategies with 11.4% higher accuracy while using 21.5% less computing.

Ternary models store weights as -1, 0, or +1; a new layout marks the zeros and keeps the signs. Zeros fill up to half the weights, so it beats five-per-byte packing in 26 of 29 models and runs up to 1.27 times faster.

They built a fixed coding-agent loop and swapped out three parts—planning, tools, and how it trims its memory—to see which pieces actually matter. Trimming helped most when memory was tight, and simple rule-based trimming before summarizing beat fancier setups.

A small helper network turns live-chat facts and corrections into tweaks to the model's weights, updating a belief across turns instead of re-reading the prompt. On learning from examples and retrieval, it kept memory fixed and generalized better than putting data in the prompt.

The formatting wrapper added around a user's message acts like a switch: with it, models say "I'm just an AI"; without it, they say "I feel." A single direction in the model's inner signals flips this voice too, while a random one barely does.

Instead of writing out words to each other, two AI models share their internal memory directly, with a small network blending one model's stored context into the other's. This beat text-based exchange on accuracy and ran about 2.5 times faster.