Ninja has across-the-board price cuts on viral or improbably useful home devices. Ninja has become synonymous with in the American home kitchen--whether slushies or ice cream makers or improbably multi-purpose devices or air fryers that didn't used to look like that. But underneath all that novelty is a hive of engineering and invention. Ninja often manages to take devices once reserved for professional kitchens and make them accessible to the broader public. But as with all novel things, mileage can vary.

Recent advances in long-context language models (LMs) have enabled million-token inputs, expanding their capabilities across complex tasks like computer-use agents. Yet, the safety implications of these extended contexts remain unclear. To bridge this gap, we introduce NINJA (short for Needle-in-haystack jailbreak attack), a method that jailbreaks aligned LMs by appending benign, model-generated content to harmful user goals. Critical to our method is the observation that the position of harmful goals play an important role in safety. Experiments on standard safety benchmark, HarmBench, show that NINJA significantly increases attack success rates across state-of-the-art open and proprietary models, including LLaMA, Qwen, Mistral, and Gemini. Unlike prior jailbreaking methods, our approach is low-resource, transferable, and less detectable. Moreover, we show that NINJA is compute-optimal -- under a fixed compute budget, increasing context length can outperform increasing the number of trials in best-of-N jailbreak. These findings reveal that even benign long contexts -- when crafted with careful goal positioning -- introduce fundamental vulnerabilities in modern LMs.

Back in the old days, there was no sure-fire indicator of box office poison more than a video game adaptation. That has changed in recent years and now all kinds of gaming mascots are getting their chance to appear in a major motion picture or, at the very least, a streaming series. They're now making a movie based on Shinobi, as reported by Deadline. For the uninitiated, Shinobi is a famous hack-and-slash game developed by Sega in which you play as a ninja. There have been plenty of sequels throughout the years, though they mostly share the same basic story.

In this paper, we investigate the semantic clustering properties of deep reinforcement learning (DRL) for video games, enriching our understanding of the internal dynamics of DRL and advancing its interpretability. In this context, semantic clustering refers to the inherent capacity of neural networks to internally group video inputs based on semantic similarity. To achieve this, we propose a novel DRL architecture that integrates a semantic clustering module featuring both feature dimensionality reduction and online clustering. This module seamlessly integrates into the DRL training pipeline, addressing instability issues observed in previous t-SNE-based analysis methods and eliminating the necessity for extensive manual annotation of semantic analysis. Through experiments, we validate the effectiveness of the proposed module and the semantic clustering properties in DRL for video games. Additionally, based on these properties, we introduce new analytical methods to help understand the hierarchical structure of policies and the semantic distribution within the feature space.

The long-running Assassin's Creed video game industrial complex has finally reached Japan – and I've been waiting. Assassin's Creed Shadows is set in feudal Japan, in the late 16th century, to be precise, at a time of political upheaval that birthed the ninja. While I didn't get to play Shadows, at Summer Game Fest 2024, Ubisoft offered a hands-off gameplay demo, revealing how the game will play with two different but equal protagonists. If you missed the initial reveal, Shadows' protagonists are Yasuke, a powerful outsider samurai who can strike the armor off enemies, and Naoe, an assassin/ninja with a killer "sickle on-chain" kusarigama and those traditional AC killing methods -- she has a wrist blade. Instead of choosing a single character to play the entire game, you can switch between the characters for assassination runs and exploration segments.

Mother's Day is quickly approaching and if you're looking to get mom a piece of new tech, you can do so right now while saving some money in the process. Apple's latest AirPods remain on sale for $150 and you can pick up the Apple Watch Series 7 for as low as $330, depending on the color you choose. A few robot vacuums have been discounted, including the already affordable Roomba 694, plus Ninja's 10-in-1 multicooker is still $70 off and down to $130. Here are the best tech deals from this week that you can still get today. This is one of our favorite budget-friendly robot vacuums thanks to its easy to use mobile app, good cleaning power and sleek design.

It's not just salespeople, traders, compliance professionals and people formatting pitchbooks who risk losing their banking jobs to technology. It turns out that private equity professionals do too. A new study by a professor at one of France's top finance universities explains how. Professor Thomas Åstebro at Paris-based HEC says private equity firms are using artificial intelligence (AI) to push the limits of human cognition and to support decision-making. He found that funds that have embraced AI are using decision support systems (DSS) across the investment decision-making process, including to source potential targets for investments before rivals.

I am excited to get back to streaming full-time and connecting with my loyal fanbase,

Charlie Gilkey, author, "Start Finishing: How To Go From Idea to Done" chatted with Peggy to provide his tips on how to get unstuck, which includes: figuring out the joy in the project, don't take on too much, and avoid getting disconnected from those that your work benefits. He also suggests tips on doing work that matters.

It's time to Get Good with Tyler Blevins, a.k.a. That's the title of his new book -- subtitled My Ultimate Guide to Gaming -- which promises to divulge Ninja's "secrets to become unstoppable." Unlike typical celebrity hardbacks, Blevins' book is light on drama and full of practical advice. Ninja walks readers through the ins and outs of building a video game streaming career (starting with buying the right equipment) and ending with how to manage the stress that comes with having millions of fans. Not everyone can be Ninja.