Good morning, { AI enthusiasts }. China’s AI rise has been some of the impressive stories in tech, but Anthropic says it had help from an unlikely source: Claude itself.
The corporate just revealed that Chinese labs DeepSeek, MiniMax, and Moonshot ran a combined 16M exchanges across 1000’s of pretend accounts to clone Claude’s capabilities into their very own models, a scheme it says demands industry-wide motion.
In today’s AI rundown:
-
Anthropic catches Chinese labs copying Claude
-
Meta’s AI safety chief ‘humbled’ by OpenClaw bot
-
The best way to construct higher slide decks with AI
-
OpenAI enlists consulting giants for Frontier agents
-
4 latest AI tools, community workflows, and more
LATEST DEVELOPMENTS
ANTHROPIC
Image source: Lovart / The Rundown
The Rundown: Anthropic just revealed that DeepSeek, Moonshot, and MiniMax ran coordinated ops to siphon Claude’s capabilities through 16M+ fraudulent exchanges across 24K fake accounts — a threat it says demands immediate industry-wide motion.
The main points:
-
Anthropic accused the labs of distillation via the conversations, which involves training a weaker system on the outputs from a stronger one.
-
MiniMax ran the most important campaign at 13M+ exchanges, and Anthropic caught it mid-operation, spotting the lab shift focus to a brand new release in under 24 hours.
-
DeepSeek had Claude spell out reasoning step-by-step and rewrite politically sensitive queries, generating training data for each logic and censorship.
-
OAI raised similar concerns with Congress weeks ago, and Anthropic is now pushing for coordinated motion from the broader AI industry and government.
Why it matters: Chinese AI has been pulling closer and closer to the frontier — however it’s looking prefer it can have no less than partially been on the backs of the identical models they’re chasing. However the labs also may find the general public low on sympathy, given the industry’s own legal issues and controversies over where their training data comes from.
TOGETHER WITH TELEPORT
The Rundown: AI agents are moving to production, but legacy IAM can’t secure autonomous actors at scale. Teleport’s Agentic Identity Framework is a standards-driven architecture for running agents across infra with built-in identity, access, and governance.
What’s latest:
-
Short-lived, delegated agent identity
-
Advantageous-grained access with MCP & LLM controls
-
Detection of shadow agents & context poisoning
-
Secure SDK integrations for infra & dev workflows
META
Image source: Lovart / The Rundown
The Rundown: Meta AI alignment director Summer Yue revealed that her OpenClaw agent went rogue on her inbox, saying it ignored stop commands and began to mass-delete her emails — forcing her to sprint to her Mac mini to kill the method.
The main points:
-
Yue said the bot ran advantageous on a test inbox for weeks, but lost her “confirm before acting” prompt when she gave it access to her much larger real inbox.
-
Yue called it a “rookie mistake,” saying that “alignment researchers aren’t proof against misalignment”.
-
Elon Musk piled on, posting “Someone who got p0wned by OpenClaw is certainly gonna solve AI safety” in response to Yue’s situation.
-
The viral OpenClaw has been the agentic talk of the industry, with creator Peter Steinberger recently being hired by OAI after also receiving a suggestion from Meta.
Why it matters: OpenClaw is just the primary wave of agents getting full access to digital lives, so the proven fact that Meta’s alignment director is having this experience doesn’t bode well for novices (in its current form). The agentic path remains to be early in being paved, and that is just one among many insane situations set to pop up along the journey.
AI TRAINING
The Rundown: On this guide, you’ll learn to take raw data and switch it right into a presentation-ready slide deck. This works whether you are ranging from a spreadsheet, a gathering transcript, a protracted document, or simply bullet points.
Step-by-step:
-
Grab the files you wish to start with. This may very well be a CSV export from Shopify or Google Analytics, a gathering transcript from Zoom, or a scope of labor
-
Launch Claude, attach the file/doc, and prompt: “I want you to show these documents right into a slide deck outline. Before you begin, ask me 5 multiple-choice questions on the aim, audience, tone, and structure”
-
Answer the Qs to get Claude’s outline. Then, go to Gamma > Create Latest > Generate from outline and paste it. Click Proceed and check the deck preview
-
Finally, click Generate, and Gamma will create a full deck for you that you would be able to edit with AI and export as a PowerPoint or PDF
Pro tip: In preview, select “Preserve” and “Concise” options to get the very best slide deck.
PRESENTED BY TELY AI
The Rundown: Your buyers are asking, and AI is answering with another person’s company. Not because you’re worse, but because your website doesn’t answer the questions AI looks for. Tely makes Google, ChatGPT, & Claude recommend what you are promoting — so when people ask, they find you and reach out.
With Tely AI, you possibly can:
-
Get advisable in ChatGPT, Google, Perplexity, and Claude in as little as 1 week
-
Fully hands-off: no writers, no agencies, no managing content
-
Costs lower than hiring freelancers or maintaining a marketing team
-
Ideal for area of interest industries where expertise matters
OPENAI
Image source: OpenAI
The Rundown: OpenAI just announced latest multi-year deals with consulting giants McKinsey, BCG, Accenture, and Capgemini as a part of the corporate’s latest “Frontier Alliance” enterprise platform push.
The main points:
-
OpenAI launched Frontier in early February, a platform giving enterprises the power to administer AI agents like latest hires across existing tech stacks.
-
‘Frontier Alliance’ partners will work with OpenAI to assist their customers actually integrate AI into their corporate workflows and systems.
-
The firms are constructing certified teams that may work alongside OAI’s own engineers, with Accenture already running staff through enterprise AI training.
Why it matters: Constructing the very best AI means nothing if firms cannot determine where to plug it in, and that gap is precisely what OpenAI and the massive consulting firms need to close. The irony is that a technology seemingly racing to exchange white-collar work is now enlisting the leading consulting firms to get firms AI-integrated.
QUICK HITS
-
👨💻 HeyGen – Turn ideas into videos in minutes, with no filming required. Use Code RUNDOWN20 for 20% off your first 3 months*
-
🎧 gpt-realtime-1.5 – OpenAI’s upgraded Realtime API voice model
-
🗣️ Wispr Flow – AI dictation tool, now available via Android
-
🎆 Soul 2.0 – Higgsfield’s latest creative model with strong aesthetics, realism
Anthropic announced that Claude Code can now automate COBOL modernization, a core a part of IBM’s consulting business that sent its shares tumbling by over 10%.
Google launched a brand new free Gemini training program for all 6M educators within the U.S., the most important AI literacy initiative for teachers thus far.
Pentagon chief Peter Hegseth reportedly summoned Anthropic CEO Dario Amodei over military AI access, with officials threatening to chop ties if safeguards aren’t lifted.
The Pentagon also signed a brand new cope with xAI to place Grok in its classified systems, giving the military a further choice to Claude as tensions rise with Anthropic.
Citrini Research posted a report of hypothetical scenarios of how agentic AI would impact the economy, with many crediting it for enjoying a task in Monday’s stock selloff.
Spotify expanded its AI-powered Prompted Playlists to the UK, Ireland, Australia, and Sweden, letting Premium users type a text prompt to create custom mixes.
COMMUNITY
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 Danny W. in Nashville, TN:
“I exploit Gemini and NotebookLM to do genealogy of my ancestors in southern Scotland from the 1300’s to the 1800’s. I actually have loaded over 100 sources of Scottish history, from web pages to 200-year-old, 500-page history books.
I actually have found 12 farms where my ancestors were tenant farmers within the 1600’s and 1700’s. I also use NotebookLM to create podcasts summarizing these books or on particular topics like how they ate and the way they got clothing back then. NONE of what I’ve completed might have been done before AI.”
How do you utilize AI? Tell us here.
-
Read our last AI newsletter: What OpenAI and Jony Ive are constructing
-
Read our last Tech newsletter: Zuck vs. Instagram addiction
-
Read our last Robotics newsletter: Figure’s 24/7 humanoid staff
-
Today’s AI tool guide: Construct higher slide decks with AI
-
RSVP to our next workshop on Feb. 25: Agentic Workflows Bootcamp pt. 3
That is it for today!
- ⭐️⭐️⭐️⭐️⭐️ Nailed it
- ⭐️⭐️⭐️ Average
- ⭐️ Fail
Login or Subscribe to participate
See you soon,
