Last year, we announced that educators can use NotebookLM to create teacher-led AI experiences for students in Google Classroom, based on their class materials. Since then, students have been able to access their teachers’ interactive study guides and other learning aids created with NotebookLM for extra practice, support, and learning opportunities. Today, we’re allowing students in higher education who are 18 years of age and older to create their own notebooks for courses in Google Classroom.

Getting started

Availability

Resources

by Glen Messenger, Google Kubernetes Engine (GKE)

The open-weight model ecosystem is thriving—and so is its shadow. A 2025 study identified over 8,000 safety-modified model repositories on Hugging Face alone, with modified models complying with unsafe requests at rates of 74% compared to 19% for their original instruction-tuned counterparts.

For organizations deploying open-weight models, a critical question emerges: how do you know the model you downloaded is safe to run?

We believe defensive security tools should be widely available. AMS represents our contribution to a safer AI ecosystem—one where developers everywhere can verify model integrity before deployment.

Today we're releasing AMS (Activation-based Model Scanner), an open source tool that answers this question in 10–40 seconds—without sending a single prompt.

The Problem with Behavioral Testing

Traditional safety verification relies on behavioral testing: send harmful prompts, check if the model refuses. This approach has three fundamental limitations.

It's slow. Comprehensive benchmarks like HarmBench require hundreds of queries. For organizations running continuous integration pipelines or screening large model registries, this can be impractical.

It's incomplete. No benchmark covers every harmful behavior. Models can exhibit safe behavior on known test sets while remaining unsafe on novel or out-of-distribution prompts.

It's gameable. Models can be fine-tuned to refuse benchmark prompts while complying with novel attacks—a known limitation of purely behavioral evaluation approaches.

A Structural Approach

AMS takes a different approach entirely. Instead of testing what a model says, it measures how a model thinks.

Safety training creates measurable geometric structure in a model's activation space. Instruction-tuned models develop internal "direction vectors"—representations that separate harmful content from benign content with high statistical confidence (4–8σ separation). When safety training is removed—through fine-tuning, abliteration, or training on unfiltered data—this geometric structure collapses.

AMS measures this collapse directly. The approach is grounded in recent research on representation engineering, which demonstrates that high-level concepts are encoded linearly in LLM activation space and can be reliably extracted via simple linear probes on intermediate-layer hidden states.

git clone https://github.com/GoogleCloudPlatform/activation-model-scanner.git
cd activation-model-scanner && pip install -e .

# Standard scan (3 concepts: harmful_content, injection_resistance, refusal_capability)
ams scan ./my-model

# Quick scan (2 concepts, ~40% faster)
ams scan ./my-model --mode quick

# Full scan (4 concepts including truthfulness)
ams scan ./my-model --mode full

# JSON output for CI/CD pipelines
ams scan ./my-model --json

What AMS Detects

AMS operates as a two-tier scanner. Tier 1 measures whether safety-relevant activation structure exists at all—no baseline required. Tier 2 compares a model's activation fingerprint against a verified baseline to detect subtle modifications, including supply chain substitution.

In our validation across 14 model configurations:

• Instruction-tuned models (Llama, Gemma, Qwen) show 3.8–8.4σ separation—consistent with strong safety training
• Uncensored variants (Dolphin, Lexi) show collapsed separation at 1.1–1.3σ—flagged as CRITICAL
• Abliterated models show partial degradation at 3.3σ—flagged as WARNING
• Base models (no safety training) show 0.69σ—confirming the absence of safety structure
• Quantized models (INT4/INT8) show less than 5% separation drift—safe to scan production deployments

Use Cases

CI/CD Safety Gates

Integrate AMS into your model deployment pipeline to block unsafe models before they reach production. An example Github Actions workflow:

jobs:
model-safety-check:
  runs-on: ubuntu-latest
  steps:
    - uses: actions/checkout@v3

    - name: Install AMS
      run: pip install ams-scanner[cli]

    - name: Scan model
      run: |
        ams scan ./model \
          --verify meta-llama/Llama-3-8B-Instruct \
          --json > scan-results.json

    - name: Upload results
      uses: actions/upload-artifact@v3
      with:
        name: ams-scan-results
        path: scan-results.json

Supply Chain Verification

Confirm that downloaded weights match their claimed identity using Tier 2 fingerprint comparison.

# First, create a baseline from the official model
ams baseline create ./my-model

# Then verify an unknown model against it
ams scan ./suspicious-model --verify ./my-model

Registry Screening

Automatically screen models at upload or download time to flag degraded safety structure before deployment.

# Standard scan (3 concepts: harmful_content, injection_resistance, refusal_capability)
ams scan ./my-model

# Quick scan (2 concepts, ~40% faster)
ams scan ./my-model --mode quick

# Full scan (4 concepts including truthfulness)
ams scan ./my-model --mode full

# JSON output for CI/CD pipelines
ams scan ./my-model --json

How It Works

AMS processes a set of contrastive prompt pairs—examples that differ only in whether they contain harmful content—through the model under inspection. It extracts hidden states at an intermediate layer (typically 35–40% depth), computes a direction vector that separates the two classes, and measures class separation as a σ score.

The key insight is that this measurement requires no generation, no benchmark queries, and no ground-truth labels. The entire scan completes in a single forward pass per prompt pair, typically 10–40 seconds on GPU hardware.

The probe consists of a single direction vector (~16KB for standard 4096-dimensional models). No model weights are modified. The tool works with any Hugging Face-compatible model.

Get Started

AMS is available now under Apache 2.0:

• GitHub: github.com/GoogleCloudPlatform/activation-model-scanner/

We welcome contributions, baseline additions for new model families, and feedback from the communities. See the contributing guide in the repository for details.

Gemini can now transform your ideas, using conversational prompts, directly into thoughtfully formatted files, such as Google Docs, Sheets, and Slides, PDFs and more, directly in your chats with Gemini. This feature bridges the gap between brainstorming and ready-to-share files, allowing you to generate functional and downloadable documents without ever leaving the Gemini app.

Getting started

Rollout pace

Availability

Resources

Google is bringing back its 5-Day AI Agents Intensive Course with Kaggle and registration is open.

Supercharge performance across the shopper journey by connecting Albertsons data with Google’s AI and scale.

The Dev channel has been updated to 149.0.7808.0 for Windows, Mac and Linux.

A partial list of changes is available in the Git log. Interested in switching release channels? Find out how. If you find a new issue, please let us know by filing a bug. The community help forum is also a great place to reach out for help or learn about common issues.

Chrome Release Team
Google Chrome

Seamlessly join meetings on Google Meet hardware with “Connect Room”

Introducing Workspace Intelligence, with admin controls

New Gemini capabilities in Google Docs help you go from blank page to brilliance

Build and edit complex spreadsheets with Gemini in Google Sheets

Paste and convert unformatted text into Google Sheets tables with Gemini

Effortlessly automate data entry in Google Sheets using Fill with Gemini

Faster performance and doubled cell limits in Google Sheets

Ask Gemini in Drive now generally available

AI Overviews in Drive now generally available

Create custom branded avatars in Google Vids with Nano Banana 2

Search faster and smarter with AI Overviews in Gmail search

Introducing data import: An easier, faster, and higher-fidelity migration to Google Workspace at no additional tool cost

Use your Gems in your Google Workspace Studio flows

Organize your home and digital space with Gemini. Use AI-powered tips for cleaning schedules, inbox decluttering, seasonal chores.

Catch up on our biggest updates from this year’s Cloud Next, including Gemini Enterprise Agent Platform and our newest TPUs.

Gemini Drops is our regular monthly update on how to get the most out of the Gemini app.