Viral AI agent molts past trademark trouble

⋅

Read Online | Sign Up | Advertise

Good morning, {{ first_name | AI enthusiasts }}. Anthropic made Clawdbot change its name, but that didn’t stop the lobster’s momentum.

The viral AI agent is now Moltbot, and it’s been taking the internet by storm from within chat apps — providing a glimpse of an agentic future that actually works, though full device access means the risks are just as real.

In today’s AI rundown:

  • Viral AI agent takes the internet by storm

  • OpenAI releases free scientific writing workspace

  • Find dozens of free AI tools with Google Labs

  • Moonshot’s K2.5 open-source model rivals frontier labs

  • 4 new AI tools, community workflows, and more

LATEST DEVELOPMENTS

AI AGENTS

🦞 Viral AI agent takes the internet by storm

Image source: Lovart / The Rundown

The Rundown: Open-source AI assistant Moltbot (formerly Clawdbot) has gone viral for its impressive agentic capabilities, operating 24/7 from within Telegram or WhatsApp — though experts warn its full system access comes with serious risks.

The details:

  • Moltbot runs locally and connects to users’ digital lives, proactively taking actions and messaging via chat apps when tasks are done.

  • The tool was renamed after Anthropic reached out over trademark concerns, with creator Peter Steinberger initially launching it as Clawdbot in December.

  • Viral demos range from negotiating and purchasing a new car to calling a restaurant via ElevenLabs after failing to book through OpenTable.

  • Many are warning of the security risks associated with full device access, including prompt injections, exposed data, and more if not properly managed.

Why it matters: Moltbot looks like a serious step up in the agentic world that actually delivers: running autonomously, keeping context across sessions, and taking real actions. But the utility comes with risk — full access to messages, credentials, and systems means a single exploit may compromise everything if not set up correctly.

TOGETHER WITH SYNK

🚩 Test your security skills live

The Rundown: Snyk’s annual Fetch the Flag CTF is back — a 24-hour virtual Capture the Flag competition hosted with NahamSec on Feb. 12th, where security pros can sharpen their skills and compete for prizes.

Here’s what you can expect:

  • 15 hands-on hacking challenges spanning web exploitation, binary, and OSINT

  • CPE credit to boost your professional development credentials

  • Rokid AR Joy 2 Glasses for top 3 teams and top 3 individual players

Register now.

OPENAI

🔬 OpenAI releases free scientific writing workspace

Image source: OpenAI

The Rundown: OpenAI just launched Prism, a free research tool that puts its most advanced reasoning model inside the scientific writing process — aiming to do for research what AI coding assistants did for software development over the past year.

The details:

  • The tool came from OpenAI’s acquisition of Crixet, a cloud-based science writing platform, which it rebuilt around its latest GPT 5.2 reasoning model.

  • Scientists can search for papers, auto-generate citations, and turn photos of whiteboard math into formatted equations, without leaving the document.

  • OpenAI revealed ChatGPT fields over 8M weekly queries on hard science topics, suggesting researchers are already leaning heavily on AI.

  • The free tool has no caps on team size or projects, a departure from typical expensive academic software pricing models.

Why it matters: Since ChatGPT’s launch, journals have been drowning in sloppy papers full of made-up citations and errors. OAI’s argument is that the problem was never AI in research — it was researchers using it blindly without context. An integrated tool like Prism could be the difference between slop and real acceleration.

AI TRAINING

🛠️ Find dozens of free AI tools with Google Labs

The Rundown: In this tutorial, you will learn how to use Google Labs’ free AI tools to build marketing content for your business. There are dozens of overlooked tools to help tackle whatever your use case may be, but we’ve selected our top three here.

Step-by-step:

  1. Navigate to labs.google/experiments to see a full list of tools.

  2. First is Pomelli. Copy your brand’s website URL and paste it into Pomelli. After ~5 minutes, you’ll get bespoke ad campaigns with image and video assets.

  3. Next, use Flow. Create start and end frames via Create Image, add them to Frames to Video with a prompt, then stitch clips using the scene builder.

  4. Finally, go to ImageFX for image iteration. Use the suggested styles to polish up your prompting skills and try locking the “seed” to keep outputs consistent.

Pro tip: Click the dropdown beside ImageFX and select MusicFX for generating music.

PRESENTED BY CONTEXTUAL AI

🤖 Introducing Agent Composer

The Rundown: Agent Composer helps teams tackle expert-level engineering tasks in high-stakes environments — compressing hours of complex routine work into minutes. Trusted by high-tech leaders like Qualcomm and Advantest, it’s built for environments where accuracy and reliability are non-negotiable.

What makes it different:

  • Unified context layer gives agents full visibility into tasks, data, and workflow context

  • Controlled agents combine dynamic intelligence with structured workflows for mission-critical reliability

  • No-code builder with pre-built templates, natural language prompts, or visual drag-and-drop

See it in action today and read the launch blog.

MOONSHOT AI

🇨🇳 Moonshot’s K2.5 open-source model rivals frontier labs

Image source: Moonshot AI

The Rundown: Chinese AI startup Moonshot just open-sourced Kimi K2.5, a new 1T-parameter model that rivals GPT-5.2 and Claude Opus 4.5 across coding, vision, and agentic benchmarks. The release includes Kimi Code, a terminal-based coding agent.

The details:

  • K2.5 tops GPT-5.2 and Opus 4.5 in on key benchmarks for agentic tasks and video reasoning, though it trails slightly on pure coding evals.

  • K2.5 shows massive cost savings over top rivals, is natively multimodal, and comes in as the top open model on Artificial Analysis’ leaderboard.

  • The model also features Agent Swarm, allowing K2.5 to manage up to 100 AI sub-agents running tasks at once across up to 1,500 steps and tools.

  • Moonshot also open-sourced Kimi Code, an agentic coding agent that works in terminals and IDEs like VSCode and Cursor.

Why it matters: The gap between open-source and closed frontier models keeps shrinking — and once again, it’s a Chinese lab leading the charge. With DeepSeek’s V4 also reportedly on deck, U.S. labs are facing pressure from multiple directions as China’s open wave keeps building at costs that are hard for top labs to compete with.

QUICK HITS

🛠️ Trending AI Tools

  • 💸 Nume – The world’s first AI CFO that every startup and SME needs. Connect to Xero and QuickBooks in 5 minutes*

  • 🦞 Moltbot (formerly Clawdbot) – Open-source personal AI assistant that lives in chat apps

  • 🚀 Kimi K2.5 – Moonshot AI’s powerful, open-source agent-focused model

  • 🎥 Ray 3.14 – Luma’s upgraded video model for professional creative workflows

*Sponsored Listing

📰 Everything else in AI today

Luma AI released Ray3.14, an upgraded video generation model that outputs native 1080p while running significantly faster and cheaper than its predecessor.

DeepSeek open-sourced OCR 2, a model for reading and extracting text from documents that tops benchmarks while being much more efficient with tokens.

AI2 introduced SERA, a new family of open-source coding agents able to be cheaply trained on private codebases with native support for Claude Code.

Google premiered ‘Dear Upstairs Neighbors’ at Sundance, an animated short made with its video AI in collaboration with artists to replicate hand-painted styles.

Alibaba’s Tongyi Lab open-sourced Z-Image, the full base version of its Z-Image Turbo, which ranked as the top open-source image model in December.

COMMUNITY

🤝 Community AI workflows

Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.

Today’s workflow comes from reader Brendan O. in Crofton, MD:

“I use Kilo Code’s Managed Indexing to find code I know exists but can’t name exactly. In large codebases, I often need to locate logic or patterns without remembering file or function names, and regular text search just doesn’t cut it.

Managed Indexing lets me describe what I’m looking for in plain English and get the right code back quickly, which makes the difference between saving an hour and losing your train of thought.”

How do you use AI? Tell us here.

🎓 Highlights: News, Guides & Events

See you soon,

Rowan, Joey, Zach, Shubham, and Jennifer — the humans behind The Rundown

Liked Liked