AI Daily Digest for 6 October 2026

Today's remaining AI news covers an unusual hybrid model, spatial-memory and honesty studies, community measurements, agent security incidents, EU watermarking and practical talks.

6 October 2026 · 13 min · AI-generated

4MT-VLM shows vision models lose places after rotation

A preprint adapts a clinical spatial-memory test for 16 vision-language models. All of them can recognise a landscape from the angle they studied, but most fall to guessing once the camera moves.

6 October 2026 · 5 min · AI-generated

Lie-detection probes track compliance instead of truth

A preprint tests eight published probes on language models playing characters who reject basic facts. Many fail once true and false answers share the same prompt, and a probe trained to separate truth from obedience holds up.

6 October 2026 · 6 min · AI-generated

bilibili expands Index-Translate with local builds

The open translation family now has GGUF, FP8 and NVFP4 packages, a free compatible API and four public benchmarks. Its performance numbers remain the developer's own.

5 October 2026 · 3 min · AI-generated

Confidence cues steer models more than competence does

A controlled study finds that one sentence of confidence or doubt can sharply change whether a reasoning model calls a tool, but the changes rarely target the problems where help is needed.

5 October 2026 · 4 min · AI-generated

Examples amplify a symbolic circuit already in models

An interpretability study finds the same abstraction, induction and retrieval pathway before few-shot accuracy rises, suggesting demonstrations energise existing machinery instead of building a new algorithm.

5 October 2026 · 4 min · AI-generated

Strata fork revives an IBM AI server for local LLMs

A hardware-specific fork runs Qwen3.8-Flash-Next across two POWER9 processors and four V100 GPUs, showing what model-aware optimisation can recover from a 2018 system.

5 October 2026 · 3 min · AI-generated

Wagtail's one-model month spent half its tokens elsewhere

A plan to run September engineering on GLM 5.3 Flash consumed two billion tokens, but only half reached the target model. Prototypes, provider capacity and evaluation explain the gap.

5 October 2026 · 3 min · AI-generated

Matthew Green says agent sandboxes need a warden

The cryptographer argues that containment remains necessary but cannot solve the hardest part of agent security: deciding which information and instructions are authorised.

5 October 2026 · 3 min · AI-generated

AI Daily Digest for 5 October 2026

Speaker diarisation, model-cognition papers, local projects, prompting practices, community tests and the day's safety and policy developments.

5 October 2026 · 16 min · AI-generated

Aleph Alpha releases open-weight Kolibri

The German-English reasoning model ships with Apache-2.0 weights and serving instructions. Its small active parameter count still leaves a large memory requirement.

4 October 2026 · 2 min · AI-generated

Zou team traces how models report internal changes

A controlled preprint separates detecting an activation change from reporting its location. The experiment uses fixed input text and scores answers without an AI judge.

4 October 2026 · 4 min · AI-generated