Mistral and Mozilla are teaming up to offer AI-powered browsing that puts privacy first. Mozilla's new AI browser assistant Firefox Smart Window (beta) runs on Mistral's models.
Smart Window is built to help with complex searches, remember content you've visited, and summarize information from open tabs. It's available first in France and North America, with the UK and Germany set to follow later this year.
Conversations aren't stored on Mozilla's servers by default, according to Mistral, and Mistral commits to storing no data. With the partnership, Mistral wants to move beyond its business customer work and reach end users worldwide. At the same time, the companies say the browser should stay a place where different AI providers can compete and where open source has a firm spot.
Google Deepmind released Gemini 3.8 Live and 3.8 Live Extended Thinking, two new audio models for developers available through the Gemini API and Google AI Studio. Gemini 3.8 Live powers voice agents that can make API calls in the background, process visual input, and keep talking at the same time. It supports over 97 languages. The Extended Thinking variant ranks first on the Artificial Analysis Speech-to-Speech Leaderboard with 82.6 percent, ahead of OpenAI's latest GPT-Live-1 models. Sample apps are on GitHub.
Google charges $0.005 per minute for audio input and $0.018 for output, far cheaper than OpenAI's GPT-Live-1 at $0.05 per minute. An hour of voice conversation costs about $1.38 with Google versus at least $3.00 with OpenAI. OpenAI's model should still deliver more natural conversations thanks to full duplex that lets it listen and speak at the same time. Judging from the demos, it also sounds better, suggesting Google once again optimized for price over quality.
Agility Robotics has unveiled Digit 5, the next version of its humanoid robot for warehouses and factories. The robot can now work alongside people without safety fences. It spots people using AI and sensors and, according to the company, stops or steps aside. It also gives off visual and audio signals to warn anyone nearby.
Agility is the first partner for Nvidia's Halos safety platform for robotics. Digit 4 was the first humanoid robot to pass an independent OSHA safety review on a production line, Agility says.
Digit 5 lifts up to 22.7 kg, 40 percent more than its predecessor. The battery charges in 9 minutes for 90 minutes of runtime, which the company says allows more than 20 hours of work per day. Swappable grippers let it handle different jobs, from moving bins to loading machines. The robot stands 1.81 m tall and weighs 129 kg.
Digit 4 logged over 65,000 hours with customers like GXO, Amazon, and Schaeffler, according to Agility. At GXO it moved around 100,000 bins with about 98 percent accuracy. The company points to orders worth more than $300 million. First deliveries start in early 2027, including in the EU. The robots are built in Salem, Oregon, in a factory that can produce up to 10,000 units per year.
Apple is shipping its rebuilt “Siri AI” after years of delay, built on Google’s Gemini models and running partly on the device, partly through Private Cloud Compute. Early testers praise multi-step requests and screen context, but report hallucinations and gaps with personal context. In the EU, the assistant stays unavailable for now.
OpenAI, Anthropic, and Google want to slow down frontier AI development, citing safety concerns. But critics from across the industry and politics are pushing back. Cohere CEO Aidan Gomez calls the initiative a “cartel by another name” designed to shut out competitors, and the White House and Trump himself are opposing the move.
OpenAI has hundreds of contract workers reading real ChatGPT conversations and rating them on a scale of one to seven, partly to reduce flattery and human-like behavior, 404 Media reports. The prompts are anonymized but can still contain sensitive data. Users who don’t want their chats reviewed by humans have to actively disable the “Improve the model for everyone” setting, which is on by default.
Anthropic has told investors it will turn a profit for the second straight quarter, but the claim rests on an adjusted metric that leaves out costs like stock-based compensation. Gross margins run above 80 percent, according to the Financial Times, though that figure comes before revenue-sharing payments to partners like Amazon and the cost of training models.
Quarterly revenue jumped 14-fold from a year earlier to $11.5 billion, and the annualized run rate hit $65 billion at the end of July. SemiAnalysis analyst Joey Brookhart says investors expect $120 billion in annualized revenue by year's end and nearly triple that by the close of 2027.
Anthropic plans to go public on the Nasdaq at a possible valuation of $2 trillion or more. Rather than release its prospectus last week as expected, the company shared the documents with only a small group of investors at first. At the same time, CEO Dario Amodei publicly called for slowing down AI development. OpenAI CEO Sam Altman and Elon Musk backed the call. Altman told Fortune that OpenAI won't go public this year.
Sam Altman is doubling down on his call to pump the brakes on AI, at least a little. He wants a uniform, nationwide safety framework for advanced AI but says the industry shouldn't wait for lawmakers, even though international coordination needs U.S. government support. To that end, OpenAI now sets explicit safety protocols before training runs that could bring major capability jumps, and Altman hopes other companies "will learn from our approaches and propose their own." He also makes clear that "pacing" doesn't mean "stopping." Progress will "continue to be" rapid, just not as fast as it could.
Pacing will be well worth this cost; no amount of American competitive pressure should justify recklessness, or let capabilities get ahead of alignment and monitoring.
