Towards AGI Launches Gen Matrix to Spotlight Generative AI Leaders and Innovators

A Thought Leadership platform to help the world navigate towards Artificial General Intelligence We are committed to navigate the path towards Artificial General Intelligence (AGI) by building a community of innovators, thinkers, and AI enthusiasts.

TheGen.AI News

Towards AGI Launches Gen Matrix to Spotlight Generative AI Leaders and Innovators

Gen Matrix has officially launched by Towards AGI, positioning itself as the ultimate resource for tracking Generative AI innovation and adoption across industries. The platform highlights the leaders shaping the future of AI, focusing on three key areas:

• Organizations: Visionaries transforming industries through Generative AI.

• Startups: Innovators redefining hardware, infrastructure, and applications.

• Leaders: Pioneers driving change and shaping the Generative AI ecosystem.

In a world where traditional thought leadership platforms struggle to keep pace with the rapid evolution of Generative AI, Gen Matrix introduces a new framework. Designed to capture both technological advancements and their real-world impacts, it provides the insights necessary to understand and navigate the progression toward Artificial General Intelligence (AGI). The much-anticipated Gen Matrix report, launching today, offers groundbreaking insights to help businesses and innovators thrive in the fast-changing AI landscape.

Readers can now access the report and nominate standout AI innovators deserving of recognition in the Gen Matrix spotlight. Gen Matrix is where AI innovation meets action. With its launch, it invites industry leaders, innovators, and businesses to join the movement in shaping the future of Generative AI.

Click here for the Matrix report.

Google Gemini 2.0: AI for the Next Generation of Search and Automation

Google has introduced Gemini 2.0, a cutting-edge AI model built for the "agentic" era, designed to enable AI agents—intelligent virtual assistants capable of performing tasks and taking actions on behalf of users. Among its key applications are Jules, an AI agent adept at independently coding, diagnosing, and resolving issues, and Project Mariner, which can execute a variety of tasks directly on the web.

The launch of Gemini 2.0 also includes Gemini 2.0 Flash, which delivers double the speed of Gemini 1.5 Pro on key benchmarks (see chart below). Notable advancements in coding capabilities are among the highlights, with further improvements expected. Developers can explore an experimental version today via AI Studio and Vertex AI, and it’s also available for trial on the @GeminiApp website, with a mobile version coming soon.

According to Sundar Pichai, Google is investing in agentic capabilities through early prototypes like Project Mariner, powered by Gemini 2.0. Mariner demonstrates the ability to understand and reason across various data types—such as pixels, text, code, images, and forms—on a browser screen, using this information to complete tasks autonomously.

Gemini 2.0 also powers AI Overviews, AI-generated summaries designed for research-intensive topics. Already reaching 1 billion users, AI Overviews will expand to more countries and languages over the next year. Sundar Pichai noted that AI has profoundly transformed Search, with AI Overviews becoming one of the platform's most popular features. Leveraging Gemini 2.0’s advanced reasoning capabilities, these summaries will soon tackle more complex topics, including advanced mathematics, multimodal queries, and coding challenges. Broader availability is expected early next year.

How AI Helped Shutterstock Cross the $100M Revenue Mark?

Shutterstock, founded in 2003 and headquartered in New York, is a global leader in licensed digital content, offering stock photos, videos, and music to creative professionals and businesses. In late 2022, the company strategically embraced generative AI, becoming one of the first stock-content providers to integrate the technology into its platform.

Transition to Generative AI

Dade Orgeron, Shutterstock’s Vice President of Innovation, has spearheaded the company's transition into generative AI. Under his leadership, Shutterstock evolved from a traditional stock-content provider into a platform offering a variety of generative AI services. Initially focused on AI-generated images, the company also launched an API for creating 3D models and plans to introduce video generation capabilities.

The Challenge and Opportunity

The emergence of mainstream generative AI tools like DALL-E, Stable Diffusion, and MidJourney in 2022 presented both a threat and an opportunity for Shutterstock. Orgeron acknowledged the disruptive potential of generative AI but saw it as a chance for innovation rather than a risk. Early adoption of the technology positioned Shutterstock as a leader in the space, even as many creative professionals were unaware of AI’s transformative impact.

Building AI Capabilities

Orgeron’s journey at Shutterstock began in 2021 with its acquisition of TurboSquid, a 3D assets company. That same year, Shutterstock acquired AI-focused companies such as Pattern89, Datasine, and Shotzr, whose expertise in data analytics laid the foundation for its aggressive move into generative AI.

Shutterstock also forged partnerships with major tech players like Meta, Alphabet, Amazon, Apple, OpenAI, Nvidia, and Reka. These collaborations, such as the partnership with Nvidia, have supported initiatives like its generative 3D services.

AI Integration and User Experience

Shutterstock prioritized accessibility in its AI tools. Its debut generative image platform, launched in January 2023, featured an intuitive web interface, making it easier for professionals to adopt compared to competitors like MidJourney, which required using Discord.

The dedicated Shutterstock.AI platform combines internally developed models with solutions from external partners, tailoring the choice of AI models to users' needs. Users can generate images via text prompts and select specific styles, such as watercolor or fish-eye photography. This multi-model approach gives Shutterstock an edge over competitors relying on single-model systems.

Addressing Risks

To mitigate potential risks to its business and contributors, Shutterstock trained all AI models using legally owned content. It also introduced a contributor fund to compensate creators whose work was included in model training. While individual creators and small businesses adopted the AI services early, enterprise clients were initially cautious, focusing on legal compliance before integrating the technology.

Results and Future Goals

By 2023, Shutterstock generated $104 million in annual revenue from AI licensing, with CEO Paul Hennessy projecting up to $250 million annually by 2027. Beyond revenue, Shutterstock values its growing partnerships with major tech companies, leveraging collaborations to develop new services and license data for AI training.

Looking ahead, Shutterstock plans to expand its AI offerings into video and 3D content, with its generative 3D API currently in beta. The company also envisions converting static images into videos as a promising future capability. Orgeron emphasized that partnerships with industry leaders like Nvidia, Meta, and HP affirm Shutterstock's strategic direction in the evolving AI landscape.

TheOpensource.AI News

New Report Highlights Security Challenges in Open Source AI Adoption

A new report titled "The State of Enterprise Open Source AI," published by Anaconda in collaboration with ETR, surveyed 100 IT decision-makers to highlight key trends in enterprise AI and open-source adoption. The findings emphasize the growing reliance on open-source components and the critical need for trusted solutions to navigate the inherent challenges of open-source AI.

Security Concerns in Open-Source AI

The report reveals that 58% of organizations incorporate open-source components in at least half of their AI/ML projects, with 34% using them in three-quarters or more. This widespread usage, however, brings significant security challenges.

"While open source tools drive innovation, they also introduce security risks that can jeopardize enterprise stability and reputation," Anaconda stated in a blog post. The report underscores the vulnerabilities organizations face and the importance of implementing robust security measures to ensure the safe deployment of AI/ML models.

Security risks emerged as a top concern, with 29% of respondents identifying them as the most pressing challenge in using open-source AI components. Issues such as vulnerability exposure, malicious code, and reliance on flawed AI insights highlight the urgent need for secure tools and trusted solutions.

Key Findings on Security Risks

The report outlines several critical security concerns:

• Vulnerability Exposure:

  • 32% reported accidental exposure of vulnerabilities.

  • 50% of these incidents were rated very or extremely significant.

• Flawed AI Insights:

  • 30% faced incorrect AI-generated outputs.

  • 23% considered the impacts very or extremely significant.

• Sensitive Information Exposure:

  • 21% reported incidents of sensitive data exposure.

  • 52% of cases had severe impacts.

• Malicious Code Installation:

  • 10% encountered accidental installation of malicious code.

  • 60% rated these incidents as very or extremely significant.

Addressing Security Challenges

The report emphasizes the need for robust security frameworks and trusted platforms to manage open-source components effectively. Anaconda highlights its platform's role in mitigating these risks by offering curated, secure open-source libraries to support innovation while maintaining stability.

Additional Insights from the Report

The report also explores broader themes in enterprise AI, including:

• Strategies for scaling AI without compromising stability.

• Methods to accelerate AI development.

• Insights on how leading organizations outpace their peers.

• Realizing ROI from AI investments.

• Overcoming challenges in fine-tuning and deploying AI models.

• Breaking down organizational silos for effective AI implementation.

This comprehensive analysis underscores the balance between leveraging the potential of open-source AI and addressing its associated risks to ensure sustainable and secure innovation.

Microsoft Unveils Phi-4: A Powerful Open-Source Small Language Model

Microsoft has unveiled its latest small language model (SLM), Phi-4, the newest addition to its open-source Phi family of foundational AI models. This release comes eight months after the debut of Phi-3 and four months after the Phi-3.5 series. The company asserts that Phi-4 offers enhanced capabilities in solving complex reasoning tasks, particularly in mathematics, while also excelling in conventional language processing tasks.

Phi-4 AI Model to Launch on Hugging Face

Unlike previous Phi models, which were accompanied by a mini variant, Phi-4 does not have a mini version. In a blog post, Microsoft announced that Phi-4 is currently available on Azure AI Foundry under the Microsoft Research Licence Agreement (MSRLA). The company plans to make the model accessible on Hugging Face next week.

Benchmark Performance

Microsoft shared internal testing benchmarks, showing that Phi-4 significantly surpasses its predecessors in capability. According to the company, Phi-4 outperforms larger models like Gemini Pro 1.5 on math competition problems. Detailed benchmark results were published in a technical paper available on arXiv.

Focus on Safety

Microsoft emphasized the robust safety features available with Phi-4 through Azure AI Foundry. These include tools to measure, mitigate, and manage AI risks throughout the development lifecycle for both traditional machine learning and generative AI applications. Enterprise users can leverage Azure AI Content Safety features, such as:

• Prompt shields for protecting input prompts.

• Groundedness detection to ensure reliability.

• Real-time monitoring for adversarial prompts, data integrity, and overall application safety.

These safety features are accessible via a single API, allowing developers to integrate them into their applications seamlessly.

Continuous Training and Efficiency

Phi-4 follows a growing trend in AI where smaller language models are trained post-deployment on synthetic data. This enables the model to quickly improve its knowledge and efficiency. However, Microsoft acknowledged that real-world outcomes of such training may vary and are not always consistent across use cases.

Phi-4 represents a significant step forward in Microsoft’s AI offerings, delivering advanced reasoning and language processing capabilities alongside robust safety measures to ensure secure deployment in enterprise applications.

TheClosedsource.AI News

ChatGPT Outage Disrupts Users Globally, OpenAI Promises Quick Fix

ChatGPT, the widely used AI chatbot, experienced a significant outage due to a technical issue, leaving millions of users unable to access the service. The disruption, which began shortly before 7 PM ET, affected not only ChatGPT but also OpenAI's API and Sora services, causing widespread problems for businesses that rely on these tools.

OpenAI Responds to Outage

OpenAI, the company behind ChatGPT, acknowledged the issue on social media, informing users that the problem had been identified and a fix was in progress. In a post on X (formerly Twitter), OpenAI stated: "We’re experiencing an outage right now. We have identified the issue and are working to roll out a fix. Sorry, and we’ll keep you updated."

Many users expressed frustration on social media, citing difficulties logging in, degraded performance, and API errors. Reports from Down Detector, a platform that tracks service disruptions, showed a significant spike in complaints about ChatGPT being offline. While the actual number of affected users may vary, the impact has been widespread.

Disruption for Businesses

The outage has caused major disruptions for companies that depend on OpenAI’s API for their projects. Many firms reported delays and performance issues, compounding the frustration among users. OpenAI has not provided an exact timeline for service restoration but assured users that they are working to resolve the issue as quickly as possible.

An OpenAI engineer posted on social media: "We have reports of API calls returning errors and login difficulties on platform.openai.com and ChatGPT. We have identified the issue and are rolling out a remediation. We are working as fast as we can to restore services and apologize for the downtime."

Meta Platforms Also Affected

In a separate incident earlier the same day, Meta's Facebook and Instagram services experienced outages, with more than 27,000 users reporting issues with Facebook and over 28,000 with Instagram. The outages, which began around 12:50 PM ET, also impacted WhatsApp, with over 1,000 users reporting problems.

These outages follow a pattern of technical issues affecting major platforms this year, including a global disruption earlier that impacted hundreds of thousands of Facebook and Instagram users for over two hours.

While OpenAI works to resolve the ChatGPT outage, the incident highlights the growing reliance on AI-powered tools and the challenges associated with maintaining uninterrupted service in a rapidly evolving digital ecosystem.

Speak Secures $78M Funding, Hits $1B Valuation with OpenAI's Backing

Speak, an AI-driven language learning platform, has secured $78 million in Series C funding, bringing its valuation to $1 billion. The round was led by Accel, with participation from existing investors, including OpenAI (via its Startup Fund), Khosla Ventures, and Y Combinator. The new funding marks a significant milestone for the company, which raised a $20 million Series B extension at a $500 million valuation just six months earlier.

AI-Powered Language Learning Inspired by Native Acquisition

Speak aims to teach languages by replicating how native speakers learn, focusing on listening and speaking. Using AI, the platform generates audio conversations and evaluates users’ spoken responses to improve fluency. Unlike traditional methods that prioritize grammar and vocabulary, Speak emphasizes practical communication skills.

The company’s approach consists of three key steps:

1. Immersing learners in listening and speaking without detailed grammatical explanations.

2. Repeated practice of phrases to internalize them without translation.

3. Anchoring new language in real-world contexts using AI.

Speak’s CEO, Connor Zwick, and CTO, Andrew Hsu, co-founded the company with the mission of helping users overcome their fear of speaking languages, particularly English. Zwick explained, “For the 1.5 billion people learning English, most already know grammar and vocabulary. The real challenge is speaking confidently.” Speak currently focuses on teaching English to users from eight native language groups, with plans to expand into Spanish and French.

Backing from OpenAI and Industry Leaders

OpenAI has been a key supporter of Speak, both as an investor and technology partner. The platform leverages OpenAI’s speech technology to power its AI-driven language learning tools. Ian Hathaway, a partner at OpenAI’s Startup Fund, stated, “We’re thrilled to see Speak’s world-class AI talent and unique product vision create transformative learning experiences for a rapidly growing global audience.”

Accel partner Ben Quazzo, who led the latest funding round, highlighted Speak’s potential as a leader in consumer AI, joining the company’s board of directors.

Speak’s Growing Reach and Consumer Offering

Although Speak has not disclosed its active user base, its app has been downloaded over 10 million times. Users typically spend 10-20 minutes daily on the platform, paying $20 per month or $99 annually—a fraction of the cost of hiring a human tutor. Speak also offers an enterprise tier, Speak for Business, which has over 200 customers.

The company remains focused on efficacy over gamification, a strategy that sets it apart from competitors like Duolingo. While gamification may be explored in the future, Zwick emphasized that Speak prioritizes real-world language proficiency.

Scaling AI-Driven Language Learning

The new funding will support Speak’s expansion into additional languages and consumer-driven features to enhance user engagement. Hsu hinted at future developments, such as creating accurate fluency assessments, while maintaining the platform’s core mission of delivering effective language learning.

With its innovative approach, high-profile partnerships, and a growing user base, Speak is positioned as a standout player in the consumer AI space, unlocking new possibilities for language education worldwide.

Don’t miss out on the insights driving the future of Artificial Intelligence! Join a community of researchers, developers, and AI enthusiasts to stay ahead of the curve in Generative AI. Each edition delivers exclusive updates, expert analysis, and thought-provoking discussions straight to your inbox. Subscribe today and be part of the journey toward AGI innovation.

Contact us for any paid collaborations and sponsorships.