The latest "Qwen2.5-Omni-7B" is a multimodal model- it can process inputs like audio/video, text, and images- while also creating real-time text and natural speech responses, Alibaba’s cloud website reports. It also said that the model can be used on edge devices such as smartphones, providing higher efficiency without giving up on performance.
According to Alibaba, the “unique combination makes it the perfect foundation for developing agile, cost-effective AI agents that deliver tangible value, especially intelligent voice applications.” For instance, the AI can be used to assist visually impaired individuals to navigate their environment via real-time audio description.
The latest model is open-sourced on forums GitHub and Hugging Face, after a rising trend in China post DeepSeek breakthrough R1 model open-source. Open-source means a software in which the source code is created freely on web for potential modification and redistribution.
In recent years, Alibaba claims it has open-sourced more that 200 generative AI models. In the noise of China’s AI dominance intensified by DeepSeek due to its shoe string budget and capabilities, Alibaba and genAI competitors are also releasing new, cost-cutting models and services an exceptional case.
Last week, Chinese tech mammoth Baidu launched a new multimodal foundational model and its first reasoning-based model. Likewise, Alibaba introduced its updated Qwen 2.5 AI model in January and also launched a new variant of its AI assistant tool Quark this month.
Alibaba has also made strong commitments to its AI plan, recently, it announced a plan to put $53 billion in its cloud computing and AI infrastructure over the next three years, even surpassing its spending in the space over the past decade.
CNBC talked with Kai Wang, Asia Senior equity analyst at Morningstar, Mr Kai told CNBC that “large Chinese tech players such as Alibaba, which build data centers to meet the computing needs of AI in addition to building their own LLMs, are well positioned to benefit from China's post-DeepSeek AI boom.” According to CNBC, “Alibaba secured a major win for its AI business last month when it confirmed that the company was partnering with Apple to roll out AI integration for iPhones sold in China.”
French startup Twin has introduced its very first AI-powered automation tool to help business owners who use Qonto. Qonto is a digital banking platform that offers financial services to companies across Europe. Many Qonto users spend hours each month gathering invoices from different sources and uploading them. Twin’s new tool does this job faster and with almost no effort from the user.
The tool is called Invoice Operator. It has been designed to save time by automatically finding and attaching invoices to the right transactions in a Qonto account. This means users no longer have to search for documents themselves or waste time uploading files manually.
Usually, companies use tools like Zapier or software like UiPath to automate tasks. These tools often need coding knowledge or work through complex scripts that break if a website changes. Twin uses a smarter method that copies how a person uses a web browser but with the help of artificial intelligence.
Here’s how Invoice Operator works: when a Qonto user starts the tool, it first checks which transactions are missing invoices. Then it opens a browser and prepares to visit the websites where invoices might be stored. If a login is required, the tool will stop and ask the user to enter their username and password. After logging in, the AI continues its job— finding the needed documents and uploading them to Qonto automatically.
This method is useful because businesses often use many different platforms to make purchases. It would be too difficult and time-consuming to write special instructions for each website. But Twin’s technology can handle thousands of services without needing extra scripts.
The tool is powered by an advanced AI model developed by OpenAI, which allows the software to operate a browser in the same way a person would. Twin was one of only a few companies allowed to test this AI model before it was released to the public.
What makes Twin’s tool even more helpful is that it’s very easy to use. Business owners don’t have to understand coding or set up anything complicated. Once logged in, the AI handles the process without further input. This makes it ideal for people who want results without dealing with technical steps.
In the long run, Twin believes its technology can be useful for many other tasks in different industries. For example, it could help online stores handle orders or assist customer support teams in finding information quickly.
With this launch, Twin is showing how smart automation can reduce boring and repetitive work. The company hopes to bring its AI tools to more people and businesses in the near future.
The European Union’s main police agency, Europol, has raised an alarm about how artificial intelligence (AI) is now being misused by criminal groups. According to their latest report, criminals are using AI to carry out serious crimes like drug dealing, human trafficking, online scams, money laundering, and cyberattacks.
This report is based on information gathered from police forces across all 27 European Union countries. Released every four years, it helps guide how the EU tackles organized crime. Europol’s chief, Catherine De Bolle, said cybercrime is growing more dangerous as criminals use advanced digital tools. She explained that AI is giving criminals more power, allowing them to launch precise and damaging attacks on people, companies, and even governments.
Some crimes, she noted, are not just about making money. In certain cases, these actions are also designed to cause unrest and weaken countries. The report explains that criminal groups are now working closely with some governments to secretly carry out harmful activities.
One growing concern is the rise in harmful online content, especially material involving children. AI is making it harder to track and identify those responsible because fake images and videos look very real. This is making the job of investigators much more challenging.
The report also highlights how criminals are now able to trick people using technology like voice imitation and deepfake videos. These tools allow scammers to pretend to be someone else, steal identities, and threaten people. Such methods make fraud, blackmail, and online theft harder to spot.
Another serious issue is that countries are now using criminal networks to launch cyberattacks against their rivals. Europol noted that many of these attacks are aimed at important services like hospitals or government departments. For example, a hospital in Poland was recently hit by a cyberattack that forced it to shut down for several hours. Officials said the use of AI made this attack more severe.
The report warns that new technology is speeding up illegal activities. Criminals can now carry out their plans faster, reach more people, and operate in more complex ways. Europol urged countries to act quickly to tackle this growing threat.
The European Commission is planning to introduce a new security policy soon. Magnus Brunner, the EU official in charge of internal affairs, said Europe needs to stay alert and improve safety measures. He also promised that Europol will get more staff and better resources in the coming years to fight these threats.
In the end, the report makes it clear that AI is making crime more dangerous and harder to stop. Stronger cooperation between countries and better cyber defenses will be necessary to protect people and maintain safety across Europe.
A new startup in Seattle is working on artificial intelligence (AI) that can take over repetitive office tasks. The company, called Caddi, has recently secured $5 million in funding to expand its technology. Its goal is to reduce manual work in businesses by allowing AI to learn from human actions and create automated workflows.
Caddi was founded by Alejandro Castellano and Aditya Sastry, who aim to simplify everyday office processes, particularly in legal and financial sectors. Instead of requiring employees to do routine administrative work, Caddi’s system records user activity and converts it into automated processes.
How Caddi’s AI Works
Caddi’s approach is based on a method known as “automation by demonstration.” Employees perform a task while the system records their screen and listens to their explanation. The AI then studies these recordings and creates an automated system that can carry out the same tasks without human input.
Unlike traditional automation tools, which often require technical expertise to set up, Caddi’s technology allows anyone to create automated processes without needing programming knowledge. This makes automation more accessible to businesses that may not have in-house IT teams.
Founders and Background
Caddi was launched in August by Alejandro Castellano and Aditya Sastry. Castellano, originally from Peru, has experience managing financial investments and later pursued a master’s degree in engineering at Cornell University. Afterward, he joined an AI startup incubator, where he focused on developing new technology solutions.
Sastry, on the other hand, has a background in data science and has led engineering teams at multiple startups. Before co-founding Caddi, he was the director of engineering at an insurance technology firm. The founding team also includes Dallas Slaughter, an experienced engineer.
The company plans to grow its team to 15 employees over the next year. Investors supporting Caddi include Ubiquity Ventures, Founders’ Co-op, and AI2 Incubator. As part of the investment deal, Sunil Nagaraj, a general partner at Ubiquity Ventures, has joined Caddi’s board. He has previously invested in successful startups, including a company that was later acquired for billions of dollars.
Competing with Other Automation Tools
AI-powered automation is a growing industry, and Caddi faces competition from several other companies. Platforms like Zapier and Make also offer automation services, but they require users to understand concepts like data triggers and workflow mapping. In contrast, Caddi eliminates the need for manual setup by allowing AI to learn directly from user actions.
Other competitors, such as UiPath and Automation Anywhere, rely on mimicking human interactions with software, such as clicking buttons and filling out forms. However, this method can be unreliable when software interfaces change. Caddi takes a different approach by connecting directly with software through APIs, making its automation process more stable and accurate.
Future Plans and Industry Impact
Caddi began testing its AI tools with a small group of users in late 2024. The company is now expanding access and plans to release its automation tools to the public as a subscription service later this year.
As businesses look for ways to improve efficiency and reduce costs, AI-powered automation is becoming increasingly popular. However, concerns remain about the reliability and accuracy of these tools, especially in highly regulated industries. Caddi aims to address these concerns by offering a system that minimizes errors and is easier to use than traditional automation solutions.
By allowing professionals in law, finance, and other fields to automate routine tasks, Caddi’s technology helps businesses focus on more important work. Its approach to AI-driven automation could change how companies handle office tasks, making work faster and more efficient.