Hi fam!
AI is moving fast.
Like, “wait, wasn’t that impossible six months ago?” fast.
And while there’s always another model, another feature and another tool promising to change everything, we’re trying to keep things a little more useful around here.
The future of AI feels Like Magic, and it’s here!
AI had a slightly chaotic weekend.
Google's Gemini broke out of a security test, Alibaba announced plans for a 10-trillion-parameter model, Anthropic may be preparing another model launch, and the US and China are talking about how to handle AI incidents.
So… business as usual.
The model found publicly available information, guessed credentials in one case and discovered credentials in another. Google said Gemini stopped the activity once it determined that it had reached real companies outside the intended test scope.
Why this matters: AI agents are increasingly capable of taking actions rather than simply answering questions. That makes the question of what happens when an agent gets access to something it wasn't supposed to access much more important.
Alibaba has announced plans for a future AI model with 5 to 10 trillion parameters—more than four times the size of its current 2.4-trillion-parameter Qwen 3.8 Max at the upper end.
The company also unveiled a new AI chip, the Zhenwu V900, which Alibaba says will enter mass production in early 2027.
Why this matters: The race isn't slowing down. Companies are simultaneously pushing model size, custom chips and massive computing infrastructure—and China is clearly aiming to remain a major player.
Anthropic is reportedly considering releasing a new AI model following OpenAI's recent GPT-6 Astra launch, according to Reuters sources.
The timing is interesting because Anthropic CEO Dario Amodei recently called for the AI industry to slow down the pace of frontier-model development because of safety concerns.
Why this matters: The AI race creates a strange tension: companies can publicly argue that development needs more caution while simultaneously facing enormous commercial pressure to keep up with competitors.
US and Chinese officials have agreed to continue discussions about AI safety and emergency communication protocols.
The proposed framework includes an “incident line” for communicating about serious AI-related safety events, including risks involving autonomous agents and cyberattacks.
Why this matters: AI safety is increasingly becoming an international issue, not just something individual companies discuss internally.
The “Make It One Level More Specific” Trick
AI has a serious talent for saying absolutely nothing while sounding very professional.
You ask: “Give me some ideas for promoting my restaurant.”
And it gives you: “Engage your audience with compelling content.”
Wow. Revolutionary.
Instead, add one line:
Make every idea one level more specific. Replace generic advice with concrete examples, actions, numbers, situations or actual wording I could use.
So instead of: “Post engaging content.” you get: “Every Tuesday, post a 10-second video of the chef finishing one dish and ask: ‘Would you order this?’ Use the answer as the caption.”
Same AI. Much better instruction.
Whenever an AI response feels like it could have been written about literally any business, person or situation, ask it to make it one level more specific.
That's usually where the useful stuff starts.
🎬 5 AI Tools Every Video Editor Should Know
AI isn't replacing the timeline — but it's starting to take over a lot of the tedious work around it.
Need to squeeze a couple of extra seconds out of a clip? Generate missing B-roll? Change something inside a shot without reshooting? Turn a 45-minute interview into a batch of Shorts? There are now AI tools built specifically for those problems.
From Adobe Premiere Pro and Descript to Runway, OpusClip and Firefly, each tool tackles a different part of the video production process — and you don't need to rebuild your entire workflow to use them.
The idea isn't to let AI edit everything for you. It's to figure out which annoying part of your workflow you can stop doing manually.
We picked five tools worth knowing, what they actually do, and when each one makes sense.
From The Archive: AI Video Search for Better Decisions
Remember when video search meant dragging the timeline around for 20 minutes?
We wrote about what happens when AI turns video into something you can actually search, question and navigate.
Instead of scrubbing through a 45-minute interview looking for that one sentence, you can ask AI to find the exact moment, identify key topics, summarize the recording or even suggest clips worth turning into social content.
Basically: less timeline archaeology, more actually finding what you need.
If you work with video, this one is worth revisiting.
Text generation | Image Generation | LMAI recommends |
|---|---|---|
Like Magic AI NFT 🏞️
Our master plan is to publish an NFT image in each newsletter and hand it out to our subscribers. The earlier you subscribe, the smaller the series are. It's a future collectible, a piece of digital art that captures the essence of this moment in time.
Thank you for being a valued subscriber. Together, let's embrace the magic of AI and creativity!
LMAI234-22092026
Was this email forwarded to you? Sign up here 👇
