As artificial intelligence accelerates its global expansion, FollowICT—a platform specializing in digital economy coverage—presents its weekly roundup of the most pressing developments shaping the AI landscape.
DuckDuckGo introduces filter to hide AI-generated images in search results
Privacy-centric browser DuckDuckGo rolled out a new setting allowing users to exclude AI-generated images from image search results. The feature, launched in response to user feedback, aims to reduce visual clutter and improve search accuracy. Users can access the toggle by navigating to the “Images” tab and selecting the new dropdown labeled “AI Images,” where they can choose to either display or hide such content.
Meta declines to sign EU’s AI Code of Practice
Meta has refused to sign the European Union’s voluntary Code of Practice for general-purpose AI models, ahead of new regulatory rules set to take effect. In a LinkedIn post, Joel Kaplan, Meta’s Head of Global Affairs, criticized the EU’s approach, claiming it poses legal uncertainties for developers and introduces requirements well beyond the scope of the bloc’s AI Act.
The EU’s framework, published earlier this month, urges companies to routinely update documentation for their AI tools, prohibits training on pirated material, and mandates honoring content owners’ requests to exclude their work from datasets.
Netflix deploys generative AI in Argentinian drama ‘El Eternauta’
Co-CEO Ted Sarandos announced that Netflix has produced its first final screen-ready AI-generated footage in the Argentinian series El Eternauta. The internal production team partnered with creators to use AI to generate a building collapse scene, which Sarandos said was completed ten times faster than traditional VFX methods and at a significantly reduced cost.
OpenAI launches multifunctional AI agent in ChatGPT
OpenAI has unveiled a general-purpose AI agent within ChatGPT, capable of autonomously executing a wide range of digital tasks. The agent can manage calendars, generate editable slideshows, and run code—all triggered by natural-language commands. Dubbed “ChatGPT Agent,” the tool integrates previous functionalities from OpenAI’s Operator and Deep Research agents.
France’s Mistral upgrades chatbot ‘Le Chat’ with Deep Search mode
French AI lab Mistral introduced several enhancements to its chatbot Le Chat, including a new “Deep Search” mode, native multilingual reasoning, and advanced image editing tools. The update follows the launch of Voxtral, Mistral’s open-source AI voice model. The new mode positions Le Chat as a more competitive AI search assistant, capable of planning, refining user queries, and summarizing results.
Swedish startup Lovable raises $200M Series A within 8 months of launch
Stockholm-based AI startup Lovable secured $200 million in a Series A funding round led by Accel, valuing the company at $1.8 billion. Similar to platforms like Cursor, Lovable enables users to build websites and apps using natural language commands. With just 45 full-time employees, the company has attracted more than 2.3 million active users. The investor lineup includes Klarna CEO Sebastian Siemiatkowski, Remote CEO Job van der Voort, Slack cofounder Stewart Butterfield, and HubSpot cofounder Dharmesh Shah.
Google rolls out AI-powered business calling feature in the U.S.
Google introduced an AI-driven business calling feature designed to contact local businesses and gather service and price information on behalf of users. The update accompanies the addition of Gemini 2.5 Pro to its search experience, enabling deeper research capabilities. Users can activate the feature by searching for terms like “pet groomers near me,” then answering a few questions to receive AI-generated insights across service providers.
Mira Murati’s Thinking Machines Lab reaches $12B valuation after seed round
Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, announced it had closed a $2 billion seed funding round led by Andreessen Horowitz. Investors include Nvidia, Accel, ServiceNow, Cisco, AMD, and Jane Street, pushing the company’s valuation to $12 billion. Despite being less than a year old, the stealth-mode lab represents one of Silicon Valley’s largest early-stage funding deals.
Mistral unveils Voxtral: first open-source voice AI model for enterprise use
Mistral released Voxtral, a voice AI model tailored for business applications. The company touts Voxtral as the first open-source system that delivers “usable speech intelligence in production,” offering an affordable alternative to proprietary platforms—claimed to be less than half the cost of competing solutions.
Meta building 5GW AI data center dubbed Hyperion
Meta CEO Mark Zuckerberg revealed that the company is constructing a massive AI data center named Hyperion, designed to provide 5 gigawatts of compute power to its superintelligence lab. The move signals Meta’s ambition to outpace rivals OpenAI and Google in the AI race. The lab is reportedly led by high-profile recruits, including the former CEOs of Scale AI and Safe Superintelligence.
OpenAI launches $50M fund to support nonprofit innovation
OpenAI announced a $50 million fund aimed at bolstering nonprofits and community organizations. The initiative follows recommendations from an internal advisory board created earlier this year. The move aligns with OpenAI’s attempt to reconcile its restructuring efforts—designed to secure major capital—with its founding mission to advance AI for public benefit.
Perplexity’s AI Browser “Comet” Could Challenge Chrome’s Dominance
Aravind Srinivas, CEO of the AI startup Perplexity, told Reuters on Friday that the company is in talks with mobile device manufacturers to pre-install its new browser, Comet, on smartphones. The move signals Perplexity’s ambition to disrupt Google’s grip on mobile browsing by leveraging user habits—most people default to pre-installed apps, which could normalize frequent use of Perplexity’s AI tools.
“It’s not easy to convince OEMs to switch the default browser from Chrome to Comet,” Srinivas said, acknowledging the inertia of user behavior on mobile platforms.
Currently in beta and available only on desktop, Comet integrates Perplexity’s proprietary AI directly into the browsing experience. Users can query personal data—emails, calendars, browsing history—and even perform tasks like scheduling meetings or summarizing webpages.
AI-Generated Band Velvet Sundown Hits One Million Streams—Then Admits It’s Not Real
Velvet Sundown released two albums and quickly amassed over a million Spotify streams. But the band later revealed that its music, promotional images, and backstory were entirely generated by artificial intelligence.
The revelation sparked debate over authenticity and transparency in music. Industry experts are now calling for streaming platforms to legally label AI-generated content, allowing listeners to make informed choices.
Initially described as a “human-guided synthetic music project,” the band denied its artificial origins. But a self-described “assistant member” later told reporters that the group used the AI music platform Suno, calling the project an “artistic hoax.”
Velvet Sundown’s official social accounts claimed the band’s identity had been hijacked, before issuing a statement confirming its AI origins: “Not entirely human. Not entirely machine. We exist somewhere in between.”
Roberto Neri, CEO of the Ivors Academy, warned that AI bands reaching mass audiences without human creators raise serious concerns about authorship and consent. “If used ethically, AI can enhance songwriting,” he said, “but right now, we’re facing deeply troubling issues.”
MIT’s New Tool Lets Anyone Train a Robot—No Coding Required
Training robots once required programming expertise. Now, engineers are designing robotic assistants that learn through demonstration.
MIT researchers have developed a 3D training interface that allows users to teach robots tasks using one of three methods: remote control, physical guidance, or live demonstration. The handheld sensor-equipped tool can be attached to common collaborative robotic arms.
Volunteers tested the device on standard factory tasks. Researchers say the tool’s flexibility could expand the pool of “robot teachers” and enable robots to learn a broader range of skills—from handling hazardous materials to packing boxes and even replicating logo designs.
Analysis
AI’s Growing Role in the Auto Industry—Market Set to Hit $186 Billion
AI-powered technologies like advanced driver-assistance systems (ADAS) and autonomous vehicles are reshaping the automotive landscape. ADAS combines AI sensors, cameras, LiDAR, and radar to enhance safety features such as adaptive cruise control, lane-keeping assistance, emergency braking, and pedestrian detection.
These systems analyze road conditions in real time and predict risks, enabling rapid decision-making and reducing accidents. Deep learning and machine learning advancements are equipping vehicles to navigate complex environments.
According to Global Market Insights, the AI automotive market was valued at $4.8 billion in 2024 and is projected to reach $186.4 billion by 2034. A McKinsey survey found that over 40% of auto executives invest up to €5 million in AI R&D, with more than 10% investing over €20 million.
At General Motors’ Factory Zero in Detroit, AI and machine learning are increasingly central. Vision systems detect maintenance issues like battery leaks and paint defects, while marketing teams use AI to analyze consumer behavior and tailor campaigns.
John Francis, GM’s head of data and analytics, said AI enables real-time production adjustments based on customer demand. “It helps us deliver the features drivers want—seamless connectivity, personalized safety, and both traditional and electric powertrains.”
AI also improves factory efficiency and preemptively resolves issues before vehicles reach customers. As White Mayhem of Northwest AI Consulting noted, AI adoption could lead to enhanced driver-assist systems. Royston Jones of Altair added that consumers now expect cars to be “more personalized, energy-efficient, and safe”—and AI will continue to deliver on those expectations.
