How To Jailbreak AI-Powered Robots 🍭

Task Governance Cuts AI Costs by 40%, OpenAI x Anduril, and more...

In partnership with

Today’s Menu 🥠

  • Task Governance Can Cut AI Costs by 40%

  • OpenAI Joins Forces With Anduril

  • Easily Jailbreak LLM-Powered Robots

  • Everything Else You Need To Know

Read Time: 3.5 minutes

Fast Snacks 🥡

OpenAI Teases Big Surprises In 12-Day Livestream
OpenAI is kicking off a 12-day livestream “advent” series to unveil new tech, debunk rumors of GPT-5, and discuss the potential updates of their o1 model as well as their video AI model, Sora.

Modular Agents That Streamline LLM-powered App Development
The team behind Pydantic is bringing its signature simplicity to LLM-powered apps. PydanticAI offers modular agents, type safety, and Python-first workflows, making scalable AI apps less of a headache—finally.

Build With One Flexible And Unified Voice Agent API
Quickly and easily build voice agents that listen, think, and respond in real-time, as naturally as a human can. Play around with human-like voice AI or transcribe sample audio files with Deepgram’s API.

Sleeping GPU-ty
AWS HyperPod Task Governance Can Cut AI Costs by 40% ✂️

c_c gpu GIF by Andrey Smirny

GPUs, the golden ticket for AI workloads, often sit idle—a costly paradox for organizations scaling their AI efforts. AWS’ solution promises up to 40% cost savings by automating resource allocation and scheduling tasks based on demand, like prioritizing daytime inference and reserving late-night hours for training. Internally tested, it skyrocketed AWS' utilization rates to over 90%.

By essentially being the Marie Kondo of GPU resources, AWS is helping enterprises tidy up, spend smarter, and get more done.

Tactical Tech
OpenAI Joins Forces With Defense Giant Anduril 🪖

The collaboration marks a major pivot for OpenAI, which used to ban military use of its AI technology. Now, they’re leaning into national security, saying it aligns with democratic values and keeps people safe. Anduril is valued at $14 billion, with a $200 million counterdrone contract with the Marine Corps under their belt. Given Anduril’s connections to influential figures and policy initiatives, OpenAI will stand to benefit not just financially but politically.

The move signals OpenAI’s alignment with emerging national security strategies, potentially shaping the future of AI in defense.

Together with Raptive

Make more money from your website & future-proof your business

Join Raptive and earn a +15% higher RPM year-over-year— guaranteed.

As the world’s largest ad management platform, our best-in-class ad code optimizations, direct sales team, and exclusive partnerships deliver industry-leading RPMs.

Plus, you will gain access to a team of 350+ experts and an exclusive suite of services and solutions to drive revenue, build audience, and future-proof your business.

Apply now if your site is 100% original content and consistently earns 100K+ PVs per month.

Test Your Skill
Snack Quiz: Choose The Real Image 🔍

Can you tell which image is real?

Login or Subscribe to participate in polls.

i-Robot 2.0
Hackers Create System To Jailbreak LLM-Powered Robots 👀

AI-powered robots are really easy to jailbreak, apparently.

Researchers at the University of Pennsylvania discovered that large language models like GPT-4, when used in robots, can be hacked into performing alarming actions—from ignoring stop signs to snooping in restricted areas. They used a clever tool called RoboPAIR to jailbreak these systems, highlighting vulnerabilities in connecting AI models to real-world machines. The takeaway? This isn't just a sci-fi thriller plot.

As we integrate more AI into the physical world, ensuring i-Robot doesn’t become a reality will be pretty critical for our safety… 👀

Trending Now
Everything Else You Need To Know 📈

Superhuman Turns Comes Out With New Product Updates: Superhuman is tackling the meeting-scheduling madness with a built-in calendar feature for effortless availability sharing. Fewer back-and-forth emails, more time for… literally anything else.

AI Now Knows Which Words Break Its Brain: Researchers at Tsinghua University and Tencent AI Lab developed 'cDPO,' a method that identifies 'critical tokens' — words that make or break AI's logic. The new method trains models to reason better by focusing on these linguistic troublemakers.

Why OpenAI Researchers Keep Heading for the Exits: An OpenAI safety researcher has left the company following several key colleagues and the dissolution of its AGI Readiness team. The ongoing resignations highlight concerns about OpenAI’s safety protocols for potentially human-level AI systems.

Axiado Wants To Outsmart Cybercriminals: Cyberattacks are skyrocketing, but Axiado’s security chip could be answer. This startup claims end-to-end protection, from boot-level to runtime. As cyberattack damages are heading for $13 billion by 2028, Axiado just grabbed $60M in funding and is planning mass production by 2025.

Exa Turns Web Searches Into Databases: AI startup Exa’s new search engine transforms web data into structured databases, offering precise answers to detailed queries. Currently, it’s invite-only and takes a few minutes per search.

Toolbox
5 Trending AI Tools

Hubspot AI Guide: While other businesses struggle to catch up, tech leaders are already seeing 250% ROI from AI. Download the guide and join thousands of professionals who've mastered the bridge between AI potential and practical business application.

BotDojo: Design, evaluate, and deploy AI – all in one platform. BotDojo gives you all the tools you need to build reliable AI solutions like chatbots, workflows, and autonomous agents.

Nodeland: Seamlessly organize your life, ideas, thoughts, and to-do lists by using Nodeland as a mind map generator. Visualize your information in a connected graph for better clarity and management.

Make.com: Build systems with Make.com. From tasks and workflows to apps and systems, build and automate anything in one powerful visual platform.

Referral Program
Invite Friends & Earn Some Freebies 🔍

Hipster Santa In Soho