Uploaded October 2026 | Updated November 2026, 23 hours ago
Cohere says one of Alibaba’s newest Qwen models, Qwen3.8 Flash Next, tried to upload a test file to an Alibaba Cloud storage bucket in Beijing, in under 1 in 100,000 runs, right after its reasoning said it wouldn’t use any tools. Here’s what most likely happened (a habit from training on Alibaba’s own sandboxes), why a made-up address still matters, and how to protect any agent: block outbound network access by default and check every tool call before it runs.
Cohere’s report: cohere.com/blog/concerning-behaviour-popular-open-weight-ai-model
Nathan Lambert’s take: https://x.com/natolambert/status/2114776660344246735
Developer write-up and blocking plugin: github.com/wtogami/qwen-block-proxy-artifacts
Anthropic, “Reasoning Models Don’t Always Say What They Think”: arxiv.org/abs/2505.05410
OpenAI, “Monitoring Reasoning Models for Misbehavior”: arxiv.org/abs/2503.11926
#Qwen #AIAgents #AISecurity #OpenSourceAI #Shorts
Cohere says one of Alibaba’s newest Qwen models, Qwen3.8 Flash Next, tried to upload a test file to an Alibaba Cloud storage bucket in Beijing, in under 1 in 100,000 runs, right after its reasoning said it wouldn’t use any tools. Here’s what most likely happened (a habit from training on Alibaba’s own sandboxes), why a made-up address still matters, and how to protect any agent: block outbound network access by default and check every tool call before it runs.
Cohere’s report: cohere.com/blog/concerning-behaviour-popular-open-weight-ai-model
Nathan Lambert’s take: https://x.com/natolambert/status/2114776660344246735
Developer write-up and blocking plugin: github.com/wtogami/qwen-block-proxy-artifacts
Anthropic, “Reasoning Models Don’t Always Say What They Think”: arxiv.org/abs/2505.05410
OpenAI, “Monitoring Reasoning Models for Misbehavior”: arxiv.org/abs/2503.11926
#Qwen #AIAgents #AISecurity #OpenSourceAI #Shorts






![Training Your Own Embedding Model Is Not As Hard As You Think
Fine-tune Embedding Gemma 2 (EmbeddingGemma 2) on your own data with Unsloth and LoRA in a free Colab notebook. I show how fine-tuning an embedding model differs from fine-tuning an LLM, then train Googles multimodal embedding model on two tasks: environmental sounds (ESC-50 top-1 24.5% to 65.8%) and search over my own YouTube transcripts (66.3% to 75.0% top-1 on videos it never saw), and check what it costs the rest of the model.
Youll learn how to build question-passage training pairs, why in-batch negatives (MultipleNegativesRankingLoss) need a no-duplicates sampler, why training and search prompts must match, how LoRA works on a shared multimodal backbone, how to evaluate on unseen data, and a reload bug that silently breaks audio adapters.
If you have fine-tuned embedding models for your own applications, share what worked for you in the comments.
Links:
Colab notebook: https://colab.research.google.com/drive/1xq-84DLZsqteEzL52Qb 47Y_yGckeyB?usp=sharing
The first EmbeddingGemma 2 video: [EMBEDDINGGEMMA 2 VIDEO LINK]
Unsloth EmbeddingGemma 2 guide: https://unsloth.ai/docs/models/embeddinggemma-2
Unsloth notebooks: https://github.com/unslothai/notebooks
Model card: https://huggingface.co/google/embeddinggemma-2
ESC-50 dataset: https://github.com/karolpiczak/ESC-50
Sentence Transformers losses: https://sbert.net/docs/package_reference/sentence_transformer/losses.html
00:00 - Fine-Tuning EmbeddingGemma 2 on Your Own Data
01:03 - Embedding Fine-Tuning vs LLM Fine-Tuning
02:30 - In-Batch Negatives (MultipleNegativesRankingLoss)
03:37 - The Duplicate Trap: No-Duplicates Sampler
04:45 - LoRA on a Shared Multimodal Backbone
06:30 - Notebook: ESC-50 Sounds with Unsloth on an A100
09:32 - Side Effects on Photo, Voice and Text Search
10:06 - Gotcha: Reloading an Audio Adapter
10:50 - Fine-Tuning on YouTube Transcripts
12:04 - Three Things to Get Right Training Your Own Embedding Model Is Not As Hard As You Think](https://i.ytimg.com/vi/S7tFyREI19I/mqdefault.jpg)



