🎉 Commercially Safe AI Video, AI Summit Tensions, Goku’s Visual Revolution, AlphaGeometry2 Tops Olympiads, Automated Trading Analysis
IP Safe Video via Adobe, AI Summit Stokes Worldwide Tensions, Goku’s Visual Leap, AlphaGeometry2 Conquers Math Olympiads, Automated AI Market Insights
Welcome to this week’s edition of AImpulse, a four point summary of the most significant advancements in the world of Artificial Intelligence followed by a cool new AI tool I’m trying out this week.
Here’s the pulse on this week’s top stories:
What’s Happening: Adobe has launched the public beta of its Firefly Video Model, promoted as the industry’s “first commercially safe AI video generation tool.” Alongside this release, the company introduced new subscription tiers starting at $9.99 for its expanded Firefly creative suite.
The details:
The new system can create 1080p video clips from either text or images, with features for fine camera direction, shot composition, and motion graphics.
Trained on licensed Adobe Stock and public domain assets, the model won’t use user-generated creations for further training.
Adobe unveiled two subscription options: Standard ($9.99/month for 20 videos) and Pro ($29.99/month for 70 videos).
Additional improvements include audio Translate and Lip Sync functions, a Scene to Image tool for referencing 3D structures, and broader integration across Adobe’s ecosystem.
Why it matters: Although many AI creative suites are on the market, few emphasize “commercial safety” as distinctly as Adobe. This focus on legally secure AI outputs could make Firefly an attractive choice for brands and agencies that need fully licensed and authorized AI-driven content at scale.
What’s Happening: The AI Action Summit in Paris ended on Tuesday, exposing rising international tensions. The United States and the United Kingdom declined to join a multinational agreement, while European leaders unveiled large-scale investment plans to rival U.S. AI dominance.
The details:
U.S. Vice President J.D. Vance cautioned against heavy AI regulations, declaring that American leadership stems from controlling chips, software, and governance.
Both the UK and the U.S. refused to sign a pledge for open, ethical AI, citing national security issues and governance disagreements.
Anthropic CEO Dario Amodei criticized the summit as a “missed opportunity,” stressing concerns over AI acceleration and the need for stronger security measures.
European Commission President Ursula von der Leyen announced a €200B AI investment plan, aiming to position Europe as an open-source alternative to U.S. leadership.
Why it matters: The summit highlighted deepening divides in AI oversight strategies. With the U.S. and the once safety-centric UK declining to sign, and China among the agreement’s supporters, it’s clear AI is turning into a pivotal global policy flashpoint capable of shifting power structures and alliances at high speed.
What’s Happening: ByteDance and the University of Hong Kong have debuted Goku and Goku+, a suite of AI models uniting image and video generation. These models claim top rankings in visual benchmarks and enable hyperrealistic commercial content for marketing and advertising.
The details:
Goku sets records across leading benchmarks for both images and videos, thanks to a unified approach that covers both outputs.
A “rectified flow” technique allows smooth transitions between stills and videos, with training conducted on 160 million images and 36 million videos.
Goku+ is specialized for marketing scenarios, generating photorealistic human avatars and product demonstrations.
The Goku+ platform provides tools for converting product photos into short video clips and creating life-like scenes of humans interacting with products for ads.
Why it matters: By producing extended, high-fidelity videos and product demos, Goku pushes AI-based content creation into new territory. This opens up fresh marketing possibilities but also brings concerns over how effortlessly AI-made content can mimic reality.
What’s Happening: Google DeepMind has introduced AlphaGeometry2, the latest math-focused AI model that solved 84% of International Mathematical Olympiad geometry problems dating back 25 years—outperforming typical gold medal winners.
The details:
The system pairs a Gemini model with a symbolic reasoning component to handle complex geometry problems needing formal proofs.
AlphaGeometry2 successfully tackled 42 of 50 problems, surpassing the average gold medalist score of 40.9 and improving significantly on its predecessor’s 54% success rate.
The training process included the creation of over 300 million synthetic theorems and proofs with progressive difficulty, leading to a far more expansive dataset than AlphaGeometry1.
Why it matters: Language models often stumble on math tasks, but DeepMind’s achievements hint that advanced AI can now conquer even highly intricate mathematical challenges. This progress could speed up breakthroughs in other math-intensive domains, including physics and engineering.
AI Tool To Know: ChatGPT’s Tasks feature can streamline your pre-market research by delivering an overview of market sentiment and technical indicators before the trading day starts.
Step-by-step:
Open ChatGPT and choose “GPT-4o with Scheduled Tasks” in the model menu.
Instruct it to run a core market performance analysis at 6:30 AM EST.
Ask for relevant news and event impact assessments to understand potential market drivers.
Request a risk evaluation along with any notable trade setups.
Pro tip: Always verify AI-generated trading suggestions with your own methods, as these models can occasionally produce inaccurate or “hallucinated” outputs.