Matthias Bastian

Dec 29, 2025

If you use a chatbot and an Urban browser extension, you might want to rethink that combination. Security researchers at Koi have found that eight browser extensions with more than eight million users combined are secretly harvesting AI conversations and potentially selling them to third parties.

Extension	Chrome	Edge
Urban VPN Proxy	6,000,000	1,323,622
1ClickVPN Proxy	600,000	36,459
Urban Browser Guard	40,000	12,624
Urban Ad Blocker	10,000	6,476

The extensions intercept conversations with ChatGPT, Claude, Gemini, Copilot, Perplexity, DeepSeek, Grok, and Meta AI. Even when the VPN is switched off, data collection continues in the background. Uninstalling the extension is the only way to stop it.

According to Koi, the data collection feature was quietly added in July 2025 through an automatic update. The data goes to Urban VPN's servers, and the privacy policy states that browsing data is shared with affiliate BiScience and that AI prompts are used for marketing analytics.

But the provider tells a different story in the Chrome Web Store, claiming data is not sold to third parties. Adding to the confusion, "Featured" badges from Google and Microsoft give users a false sense of security. Urban Ad Blocker for Edge is the only extension without a Featured badge.

Comment Source: Koi

Matthias Bastian

Dec 29, 2025

Short News

Zara is using artificial intelligence to digitally edit photos of models. The Spanish fashion giant reaches out to models and asks for permission to reuse existing images, reports City AM from London. Using AI, the company then dresses the models in new clothes and places them in different settings, eliminating the need for new photo shoots. Two models told City AM they receive the same payment as they would for an actual shoot. But everyone else who would normally be involved in a photo shoot, from makeup artists to photographers and stylists, likely gets nothing.

Zara says AI will supplement traditional photo shoots, not replace them. The shift comes during a rough patch for the retailer, with UK sales hitting their lowest point in six months this past November. Competitors H&M and Zalando announced similar plans over the summer, creating AI-based "digital twins" of models.

Comment Source: CityAM

Matthias Bastian

Dec 29, 2025

Short News

The Association of Chartered Certified Accountants (ACCA), the world's largest accountancy organization, will stop offering online exams starting March 2026. The decision comes as AI-powered cheating outpaces detection methods. "We're seeing the sophistication of [cheating] systems outpacing what can be put in, [in] terms of safeguards," ACCA CEO Helen Brand told the Financial Times.

The organization serves nearly 260,000 members and more than 500,000 students. Despite efforts to combat cheating, "people who want to do bad things are probably working at a quicker pace," Brand said. One student told the FT that a friend photographed exam questions and fed them into an AI chatbot. A recent study found that current reasoning models can pass the demanding CFA certification.

Online exams launched during the pandemic. The ACCA is now overhauling its qualification for the first time in a decade, adding focus on AI, blockchain, and data science. The ICAEW, another major accountancy body, still offers some online exams but reports rising fraud cases.

Comment Source: Financial Times

Matthias Bastian

Dec 29, 2025

Short News

The Creators Coalition on AI (CCAI) brings together filmmakers and artists pushing for common standards around AI use in the entertainment industry. Oscar winners Daniel Kwan and Jonathan Wang founded the group alongside actor Joseph Gordon-Levitt. Natalie Portman, Cate Blanchett, Paul McCartney, Guillermo Del Toro, and Mark Ruffalo have signed on as supporters.

The coalition isn't against AI outright—instead, it's pushing for clear guidelines. The group focuses on four main issues: getting consent and fair pay when creative work is used for training data, protecting jobs in the industry, establishing safeguards against deepfakes and misuse, and keeping human creativity at the center of the creative process. The CCAI aims to become a central hub for industry-wide conversations about these issues.

Comment Source: CCAI

Matthias Bastian

Dec 29, 2025

AI and society

One in five YouTube Shorts shown to new users is AI-generated slop, study finds

A recent study shows that low-quality AI videos are already a million-dollar industry firmly embedded in YouTube’s daily feed. Set up a fresh account today, and one in five Shorts will be AI slop. And the technology is just getting started.

Read full article

Comment

Matthias Bastian

Dec 28, 2025

AI and society

OpenAI's Sam Altman warned of AI's "superhuman persuasion" in 2023, and 2025 proved him right

Matthias Bastian

Dec 28, 2025

Short News

In case there was any doubt that Nvidia's Groq deal is anything but a takeover in disguise: according to Axios, roughly 90 percent of the workforce—including CEO Jonathan Ross and President Sunny Madra—is moving to Nvidia. Groq will continue as an independent company under new CEO Simon Edwards.

Though officially a non-exclusive license agreement worth around $20 billion, employees and shareholders are walking away with significant payouts. Staff moving to Nvidia get cash for vested shares and Nvidia stock for unvested ones; even those at Groq for less than a year will have their vesting cliff waived for immediate liquidity. Shareholders receive about 85 percent upfront, another 10 percent in mid-2026, and the rest by year's end.

Since 2016, Groq has raised around $3.3 billion from investors including Blackrock, Samsung, and Social Capital. They're now seeing substantial returns, as the deal pushed the startup's valuation from $7 billion to roughly $20 billion. For a more in-depth look at why Nvidia made this move, see my analysis.

Comment Source: Axios

Matthias Bastian

Dec 28, 2025

AI in practice

Microsoft CEO Nadella tells managers Copilot's Gmail and Outlook integrations ‘don't really work’ and steps in to fix them

Microsoft CEO Satya Nadella reportedly called Copilot’s Gmail and Outlook integrations “not smart” and is now personally stepping into product development. The worry: despite its strong starting position in AI software, Microsoft is falling behind.

Read full article

Comment

Matthias Bastian

Dec 27, 2025

Short News

OpenAI is hiring a Head of Preparedness. The position focuses on safety risks posed by AI models. OpenAI CEO Sam Altman points to the now well-documented effects of AI models on mental health as one example. Beyond that, the models have become so capable at cybersecurity that they can find critical vulnerabilities on their own.

This is a critical role at an important time; models are improving quickly and are now capable of many great things, but they are also starting to present some real challenges.

One of the key challenges for the new leader will be making sure cybersecurity defenders can use the latest AI capabilities while keeping attackers locked out. The role also covers safe handling of biological capabilities—meaning how AI models release biological knowledge—and self-improving systems.

OpenAI has faced criticism recently, particularly from former employees, for neglecting model safety in favor of shipping products. Many safety researchers have left the company.

Comment Source: Altman via X | Job listing

Matthias Bastian

Dec 27, 2025

Short News

AI startup Resemble AI is taking on Elevenlabs with "Chatterbox Turbo," an open text-to-speech model that can clone voices from just five seconds of audio. The company claims its new model beats both Elevenlabs and Cartesia on voice quality while delivering first audio output in under 150 milliseconds. That speed could make it attractive for developers building real-time agents, customer support systems, games, avatars, and social platforms. Companies in regulated industries might also find the model's built-in "PerTh" watermark useful for verifying that speech was AI-generated.

Resemble AI released Chatterbox Turbo under an MIT license, meaning anyone can use, tweak, and redistribute it for free, even for commercial projects. The model is available to try on Hugging Face, RunPod, Modal, Replicate, and Fal, with the full code available on GitHub. Resemble AI also offers a hosted service, with a low-latency version on the way.

Comment Source: Dev Shah via X