Last Week’s AI News #22

The past few days in the AI world have brought a flood of exciting news and innovations. But if you think the buzz around PHOENIX-4 or GEMINI 3.1 PRO was the most impressive… well, allow us to politely disagree. 

Because the real (okay, okay… at least in our galaxy, the real) HUGE news is something else entirely…

Let’s take a quick spin through the news and see what kind of big vibes we’re serving:

• ChangeAI got a fresh new look
• AI Mastery Program: Do A Week's Work In A Single Day
• Strategic AI Accelerator: Increase Your Team’s Output
• ChangeAI featured in Verslo Žinios, Lithuania’s largest business news portal
• Google brings Lyria 3 to Gemini, AI music creation goes mainstream
• Tavus launches Phoenix-4, real-time AI avatars that actually feel human
• Altman and Amodei’s awkward summit moment highlights AI’s growing rivalry
• Google launches Gemini 3.1 Pro, back in the world’s top model race
• Anthropic accuses Chinese labs of massive Claude distillation scheme
• Everything else that happened in AI last week

CHANGEAI GOT A FRESH NEW LOOK

Yes, this is one of our biggest updates, a complete refresh of our website. The brand has taken on a new form, and our services have become even more precise and effective.

Every new step reflects our commitment to improving the experience and creating real value.

The new website is just the beginning.

AI MASTERY PROGRAM 

In just 4 weeks, go from AI noise to fully working assistants, automations, and a clear plan to optimize your work, and your team’s, using Google or Microsoft AI tools.

Who it’s for: Professionals looking to fundamentally transform their work routines.
Outcome: Save at least 5 hours per week, confidently apply AI tools across workflows, and learn to spot processes ripe for AI automation.

Price: €559, get 10% off with code KOVAS (valid until March 1). Register here.

STRATEGIC AI ACCELERATOR

The AI Mastery Program, supercharged: two intensive sessions plus homework, tailored to your company’s needs, with deep dives into Google and Microsoft AI ecosystems.

Who it’s for: Teams of 8 or more.
Outcome: Unified AI knowledge and practical skills, save at least 5 hours per participant per week, confidently apply AI tools across workflows, learn to spot automation opportunities, and define clear AI transformation directions for your company.

Price: €4000,  get 10% off with code KOVAS (valid when booked by March 1). Register here. 

CHANGEAI FEATURED IN VERSLO ŽINIOS, LITHUANIA’S LARGEST BUSINESS NEWS PORTAL

ChangeAI’s AI Iceberg framework was recently featured in Verslo žinios, Lithuania’s largest business news portal.

Odeta, founder and CEO of Change AI, explains that the reality in most companies is similar: while some employees are still “plowing with the old tools” and using outdated methods, others are already “driving the AI tractor,” enabling them to accomplish far more.

Read more...

GOOGLE BRINGS LYRIA 3 TO GEMINI, AI MUSIC CREATION GOES MAINSTREAM

Google has launched Lyria 3 inside Gemini, allowing users to generate 30-second music tracks, complete with lyrics and cover art, from a simple text prompt, photo, or video.

The model automatically handles genre, tempo, vocals, and lyrics. Every track is marked with Google’s SynthID watermark, and users can also check if uploaded audio was AI-generated.

Creators on YouTube can access Lyria 3 via Dream Track for Shorts, making custom soundtracks easier than ever.

Why does it matter for businesses?

AI music is no longer niche, it’s becoming mainstream and instantly accessible. Businesses can create custom soundtracks for marketing, social media, and branded content in minutes, cutting production time and costs while enabling faster, more personalized campaigns at scale.

TAVUS LAUNCHES PHOENIX-4, REAL-TIME AI AVATARS THAT ACTUALLY FEEL HUMAN

Tavus has introduced Phoenix-4, a real-time human rendering model that generates AI avatars with dynamic facial expressions, emotional shifts mid-conversation, and contextual listening reactions.

The system renders every frame from scratch, supports 10+ emotional states, and runs in HD at 40 FPS for smooth live video interactions. Tavus positions the tech for healthcare, education, and sales, where feeling “heard” can directly impact outcomes.

Why does it matter for businesses?

AI avatars are moving beyond robotic scripts to emotionally responsive, real-time interactions. That means more engaging sales calls, scalable customer support, personalized training, and digital healthcare experiences that feel genuinely human, reducing costs while increasing connection. But as realism improves, so do the risks of misuse, making trust and verification equally critical.

ALTMAN AND AMODEI’S AWKWARD SUMMIT MOMENT HIGHLIGHTS AI’S GROWING RIVALRY

At the India AI Impact Summit, Sam Altman and Dario Amodei appeared to avoid linking hands during a staged photo with Narendra Modi, a viral moment symbolizing tensions between OpenAI and Anthropic.

The awkward exchange follows recent public jabs, including Anthropic’s Super Bowl ad criticizing ChatGPT ads and OpenAI’s hiring moves that may have added fuel to the rivalry.

Why does it matter for businesses?

AI is becoming critical infrastructure for modern companies, and visible rivalry between top labs signals increasing competition, faster product cycles, and potential fragmentation of ecosystems. For businesses, this means more innovation and choice, but also greater strategic risk: platform lock-in, shifting standards, and geopolitical influence are now part of the AI equation.

GOOGLE LAUNCHES GEMINI 3.1 PRO, BACK IN THE WORLD’S TOP MODEL RACE

Google has released Gemini 3.1 Pro, delivering a major reasoning upgrade and benchmark-leading performance, without raising API prices.

The model scored 77.1% on ARC-AGI-2, outperforming frontier competitors from Anthropic and OpenAI. It also leads in science, coding, and agentic search benchmarks.

Google positions 3.1 Pro as the intelligence core behind its recent Deep Think update, now integrated across the Gemini ecosystem.

Why does it matter for businesses?

Frontier-level AI performance is getting stronger, without getting more expensive. For businesses, that means better reasoning, coding, research, and automation capabilities at the same cost, making advanced AI more accessible for large-scale deployment. As competition intensifies among top labs, companies benefit from faster innovation cycles, stronger models, and improved pricing leverage.

ANTHROPIC ACCUSES CHINESE LABS OF MASSIVE CLAUDE DISTILLATION SCHEME

Anthropic revealed that DeepSeek, Moonshot AI, and MiniMax allegedly ran coordinated operations involving 16M+ fraudulent interactions across 24K fake accounts to extract capabilities from its Claude model.

Anthropic claims the labs used distillation tactics, training weaker systems on outputs generated by stronger ones, and is calling for industry-wide and government action. OpenAI previously raised similar concerns with U.S. lawmakers.

Why does it matter for businesses?

As AI becomes strategic infrastructure, model security and intellectual property protection are turning into major competitive and geopolitical issues. If frontier capabilities can be replicated through large-scale distillation, it reshapes the economics of AI development, accelerating global competition while increasing regulatory scrutiny. For businesses, this signals rising risks around data security, compliance, and ecosystem stability, especially when operating across international markets.

EVERYTHING ELSE THAT HAPPENED IN AI LAST WEEK

• Anthropic’s Claude Code revealed that last month “100% of contributions” to the agentic tool were written by Claude Code itself. This shows AI’s growing ability to self-generate useful code autonomously.
• Liquid AI released LFM2-2.6B-Exp, a tiny on-device model that performs well in math, instruction following, and knowledge benchmarks,ideal for SMBs wanting local, lightweight AI tools.
• Alibaba introduced MAI-UI, an AI agent that can autonomously control smartphone apps and complete multi-step tasks on mobile devices, streamlining repetitive business workflows.
• Adobe partnered with AI video startup Runway, integrating models like Gen-4.5 into Adobe Firefly AI studio, enabling SMBs to produce professional video content faster and cheaper.
• Google launched a free Gemini training program for all 6M educators in the U.S., the largest AI literacy initiative for teachers, a model for SMBs seeking internal AI upskilling.
• Spotify expanded AI-powered Prompted Playlists to the UK, Ireland, Australia, and Sweden, letting Premium users create custom mixes from text prompts, an example of personalized AI-driven customer engagement.
• SoftBank is acquiring DigitalBridge for $4B, adding data center and digital infrastructure to its AI strategy, highlighting opportunities for SMBs to leverage cloud and AI infrastructure partnerships.
  
• OpenAI acqui-hired enterprise AI search startup Nerve to scale ChatGPT search capabilities, signaling new opportunities for SMBs to access enterprise-grade AI search tools.
• Tencent open-sourced Hunyuan Motion 1.0, a 1B parameter model that generates 3D character animations from text prompts, perfect for small studios or marketing teams creating 3D content.
• Anthropic announced Claude Code can now automate COBOL modernization, a core IBM consulting task, showing AI’s ability to handle legacy business processes efficiently.

AI is evolving faster than ever, and staying informed is key to staying ahead. If you want to keep up with the latest updates, insights, and strategies from ChangeAI, subscribe to our newsletter and never miss a beat.