As artificial intelligence continues to reshape industries at breakneck speed, FollowICT presents a weekly roundup of the most consequential developments in the AI landscape—from corporate shakeups to geopolitical tensions.
OpenAI’s GPT-5 Pricing Sparks Industry-Wide Shockwaves
OpenAI stunned the tech world for the second time in a week with the release of GPT-5, its latest flagship model, just days after unveiling two open-source models. CEO Sam Altman declared GPT-5 “the best model in the world,” citing its exceptional performance across a wide range of tasks—especially programming.
But it’s the pricing that’s turning heads. Altman tweeted, “I’m very happy with the price we’re offering!” GPT-5’s API costs just $1.25 per million input tokens and $10 per million output tokens, with buffered inputs priced at $0.125 per million. That undercuts Google’s Gemini 2.5 Pro, which charges more for high-volume usage beyond 200,000 requests.
The move also places pressure on Anthropic’s Claude Opus 4.1, which starts at $15 per million input tokens and $75 per million output tokens—though Anthropic offers steep discounts for prompt caching and batch processing.
Meta Acquires AI Audio Startup WaveForms
Meta has acquired WaveForms, a fast-growing AI audio startup, for an undisclosed sum, according to The Information. The deal bolsters Meta’s newly formed Superintelligence Labs and marks its second major AI audio acquisition in a month, following its purchase of PlayAI.
Founded just eight months ago, WaveForms raised $40 million from Andreessen Horowitz in a round that valued the company at $160 million pre-money, per PitchBook data.
Google Revamps Finance Platform with AI-Powered Features
Google announced a sweeping upgrade to its Google Finance tool on Friday, integrating AI to deliver real-time financial insights and interactive charting. Users can now ask complex financial questions and receive AI-generated answers with embedded links to relevant sources.
The new charting tools go beyond asset performance, allowing users to visualize technical indicators like moving average envelopes and toggle between candlestick and other chart formats.
Google’s Gemini Launches “Guided Learning” to Rival ChatGPT’s Study Mode
With the academic year approaching, Google unveiled “Guided Learning,” a new feature within Gemini designed to foster deeper understanding rather than surface-level answers. The launch comes just one week after OpenAI introduced “Study Mode” for ChatGPT, aimed at promoting critical thinking.
Both companies appear to be responding to growing concerns that AI chatbots may undermine learning by offering direct answers. Instead, these new tools position chatbots as educational companions.
Gemini’s Guided Learning breaks down problems step-by-step and adapts explanations to user needs, using visuals, videos, and interactive quizzes to reinforce understanding.
Tavily Raises $20M to Build AI Agents with Real-Time Web Access
AI startup Tavily has secured $20 million in Series A funding led by Insight Partners and Alpha Wave Global. The latter, known for backing SpaceX, X (formerly Twitter), Monzo, and Hyperloop, also led Tavily’s previous round, bringing total funding to $25 million.
Tavily’s platform enables AI agents to browse live web data through a unified search layer. Since launch, the company has seen downloads surge past one million per month and claims to serve Fortune 500 firms and fast-growing tech companies.
OpenAI Releases Two Open-Weight Reasoning Models
On Tuesday, OpenAI released two open-weight AI reasoning models with capabilities comparable to its O-series. Available for free on Hugging Face, the models are described as “state-of-the-art” across multiple open model benchmarks.
The larger model, gpt-oss-120b, runs on a single Nvidia GPU, while the lighter gpt-oss-20b can operate on a consumer laptop with 16GB of RAM.
Google’s NotebookLM Opens Access to Younger Users
Google’s AI-powered note-taking app, NotebookLM, is now available to users aged 13 and up, expanding beyond its previous 18+ restriction. The app is also accessible to all Google Workspace for Education users.
The move aims to help younger students better understand their coursework. Features include podcast-style audio summaries, interactive mind maps, and visual idea summaries. NotebookLM recently added video snapshots that convert notes, PDFs, and images into dynamic presentations.
The expansion comes amid rising concerns over AI in education, particularly around data privacy and misuse. Google says it enforces stricter content policies for users under 18 and does not review user conversations or use them for AI training.
ElevenLabs Debuts AI Music Generator for Commercial Use
ElevenLabs, a leader in AI voice synthesis, has launched a new model that allows users to generate music—cleared for commercial use. The release marks a strategic expansion beyond its core focus on text-to-speech tools.
Over the past three years, ElevenLabs has become a dominant force in voice AI, branching into chatbots and speech translation. The music generator adds another dimension to its growing suite of creative tools.
South Korea Mobilizes National AI Model to Compete with U.S. and China
South Korea has tasked its largest corporations and most promising startups with building a national foundational AI model using primarily domestic technologies. The initiative aims to establish a self-sufficient AI industry and position Seoul as a viable alternative to U.S. and Chinese dominance.
The Ministry of Science and ICT announced five consortiums selected to develop the models, which will incorporate Korean-made semiconductors and software.
U.S. Senators Raise Alarm Over Chinese AI Models and Data Security
A group of seven Republican senators, led by Sen. Ted Budd, has called on the Department of Commerce to investigate potential data security risks posed by Chinese open-source AI models, including those from startup DeepSeek.
The lawmakers are urging the department to assess whether these models transmit sensitive personal or institutional data to Chinese servers, and whether they are being used by the Chinese military or affiliated entities.
The letter also requests details on any unauthorized access to export-controlled semiconductors or violations of U.S. model usage terms.
DeepMind Unveils “Genie 3,” a Step Toward Artificial General Intelligence
Google’s DeepMind division introduced Genie 3 on Tuesday, a new foundational model designed to simulate interactive worlds for training multi-purpose AI agents. Researchers say the model represents a critical milestone toward artificial general intelligence (AGI).
“Genie 3 is the first real-time interactive world model for general use,” said Shlomi Fruchter, DeepMind’s director of research, during a press briefing. “It goes beyond previous limited models. It’s not tied to any specific environment and can generate photorealistic or fantastical worlds—and everything in between,” he told TechCrunch.
Still in research preview, Genie 3 builds on its predecessor Genie 2 and DeepMind’s latest video generation model, Veo 3, which is said to possess a deep understanding of physics.
Xiaomi Launches Open-Source AI Voice Model for Smart Devices and Vehicles
Chinese tech giant Xiaomi has unveiled an open-source AI voice model designed to power its smart home and automotive technologies. The model, named MiDashengLM-7B, marks Xiaomi’s entry into the competitive race to develop multimodal AI tools that go beyond text-based applications.
MiDashengLM-7B builds on Xiaomi’s foundational voice model, already integrated into its smart vehicles and home appliances. The release signals the company’s intent to deepen its AI capabilities across consumer electronics.
Google’s Genie 3 Could Train AI Robots in Virtual Warehouses
Google is positioning its Genie 3 model as a critical step toward artificial general intelligence (AGI), enabling AI systems to interact with realistic simulations of the physical world.
According to Google, Genie 3 can be used to train robots and autonomous vehicles in virtual environments that mimic real-world settings such as warehouses. DeepMind, Google’s AI research division, claims that “world models” like Genie 3 are essential to achieving AGI—a theoretical level of intelligence where systems can perform most tasks on par with humans, rather than excelling at narrow functions like chess or translation.
“We expect this technology to play a pivotal role in our pursuit of AGI, and for agents to take on a larger role in the world,” DeepMind said in a statement quoted by The Guardian.
Beijing Store Sells Humanoid Robots from Over 40 Chinese Brands
A newly opened store in Beijing is selling more than 100 humanoid robots from over 40 Chinese brands, including UBTECH Robotics and Unitree Robotics. The store, which opened Friday, is among the first in China to offer consumer-facing humanoid robots, reflecting the country’s ambition to lead in AI and robotics.
Analysis | How Ambitious Entrepreneurs Can Harness AI to Scale Their Startups
Executives across industries agree: AI is transforming workflows, redefining business models, and accelerating innovation at an unprecedented pace. In a global survey by Deloitte, 78% of executives said they plan to increase AI investments over the next year.
But the revolution isn’t limited to large enterprises. Generative AI tools are unlocking capabilities once reserved for big teams and hefty budgets. Still, adoption among small and medium-sized businesses remains limited. According to Harvard Business Review, only 21% of small businesses currently use or plan to use AI in the next two years.
Yet GEM data from Harvard reveals a more nuanced picture. Ambitious entrepreneurs—those expecting to hire at least 20 new employees in the next five years—are far more bullish. A striking 87% believe AI will be critical to their business model and strategy within three years, and over 90% anticipate positive impacts on revenue and growth. Key benefits cited include product innovation, improved efficiency, better customer targeting, and enhanced risk management.
To guide high-growth entrepreneurs, Harvard developed a flexible framework for AI adoption. It’s also useful for investors, advisors, and ecosystem partners. The framework emphasizes tailoring AI strategies to each startup’s resources, context, and aspirations. A tech-heavy startup will approach AI differently than a lean, bootstrapped venture. But one principle holds: AI adoption must be bottom-up, not just top-down.
Set the Pace and Direction of AI Adoption
The first question isn’t whether to adopt AI—but how, and how fast. While some startups aim for sweeping applications, most benefit from gradual, experiment-driven adoption. Just as entrepreneurs build minimum viable products, they can develop minimum viable use cases for AI.
Early applications like robotic process automation (RPA) or AI agents for repetitive tasks rarely require major operational overhauls. These low-risk pilots build momentum and internal support.
Some startups prioritize high-value, revenue-generating opportunities—like using AI to launch new products or expand customer reach. Others begin with lightweight, cost-effective tools embedded in existing systems.
Importantly, entrepreneurs don’t need in-house AI expertise from day one. Many start by partnering with companies that offer AI-infused tools tailored to their sector. For example, Netic provides AI solutions for HVAC, plumbing, and electrical services—automating customer interactions, appointment scheduling, and marketing. These tools help small service businesses boost efficiency without expanding headcount or infrastructure.
Such tools offer entrepreneurs a low-risk way to test AI’s potential. For founders juggling multiple roles, they free up time to focus on growth. While outsourcing AI may not yield immediate competitive advantage, it builds organizational confidence and paves the way for deeper internal experimentation.
Strengthen the Human-AI Partnership
AI excels at data processing, pattern recognition, and rule-based tasks—but it can’t replicate human judgment, empathy, or adaptability. The opportunity lies in designing workflows that leverage both machine precision and human creativity.
Consider how service-oriented startups use AI to personalize fashion. Style DNA began as a personal styling service helping users optimize their wardrobes. To scale and enhance customization, it integrated AI into its platform—using image recognition and preference learning to suggest outfits based on user-uploaded photos and style profiles. While the app offers data-driven recommendations, users retain control over final selections. This hybrid model blends human taste with machine intelligence and promotes sustainable fashion by encouraging reuse.
In startups where employees wear multiple hats, AI can shift team members from repetitive tasks to creative, customer-facing roles—boosting engagement, retention, and company value. Entrepreneurs should view AI not as a threat, but as a catalyst for human potential.
Empower Employee-Led AI Adoption
Unlike large corporations, most startups can’t afford dedicated AI teams. But they can still lead adoption by empowering “citizen developers”—tech-savvy, curious employees willing to experiment without formal AI training. These team members understand company workflows and culture, making them ideal early adopters.
Today, nearly any language model can generate code. Specialized tools can build websites, presentations, business plans, market research, product designs, logos, and more—often at little or no cost. In startup environments, this accessibility is a force multiplier.
Bottom-up adoption builds trust. GEM data shows entrepreneurs worry their employees fear AI—but peer-led initiatives are more likely to gain traction than top-down mandates. Over time, these early adopters become internal champions, modeling best practices and building AI fluency across the organization.
Entrepreneurs should nurture this momentum with targeted development, recognition, and clear incentives. Done right, employee-led AI adoption not only boosts productivity—it fosters a more resilient, innovative culture.
