Google announced on Friday that users can now remove the visible watermark from its AI generations, covering images, videos, and songs. The change applies to content made with Nano Banana, Omni, and Lyria, and the toggle is available in Gemini and Flow, Google’s video editor, with support in Search coming soon.

The important detail is what does not change. Turning off the visible mark has no effect on the invisible SynthID watermark or the C2PA metadata that identifies AI-generated content. Josh Woodward, Google’s VP for Gemini, framed it as a balance between creative control and safety, saying visible watermarks are now optional while the invisible layers remain in place, so anyone can still use Gemini or Search to check whether an image was AI-generated

A visible watermark stamped across an image or video makes it much harder to use for professional or creative work, and creators have been pushing back on that for a while. At the same time, the need to identify AI-generated content has not gone anywhere, which is why Google is keeping the detection layer intact and just moving it out of sight. Once the feature reaches your account, the setting sits under Settings, then Media Watermark.

Google is also open-sourcing a new library called Credentio, which lets developers build local C2PA validation directly into their own apps. That is a meaningful addition, since detection tools only matter if they are widely available rather than locked inside one company’s products.

It arrives right after Anthropic’s much-debated decision to watermark text and files generated by Claude in order to comply with EU regulations, a move that drew a lot of frustration from users. Google is going in the opposite direction on visibility while keeping the underlying transparency infrastructure in place, which is arguably the more sustainable version of the same idea.

OpenAI has rolled out a new mode called Ultrafast, built to dramatically speed up how quickly GPT-5.6 Sol, its most powerful model, gets work done. The company says it runs at 14 times the speed of standard processing, delivering up to 750 output tokens per second.

The reason this matters is the trade-off it removes. Until now, getting real-time speed out of an AI model usually meant switching to a smaller or more specialised version and accepting weaker results. OpenAI is framing Ultrafast as a different direction entirely, describing it as more useful work per second rather than less capable work delivered faster.

The target audience here is clearly enterprise. OpenAI points to workflows where speed genuinely changes the outcome, including incident response, customer service and support, financial market analysis, and e-commerce. In situations where a delay of several seconds actually costs something, a 14x speedup is a meaningful selling point rather than a nice-to-have.

The technology behind it comes from OpenAI's partnership with chipmaker Cerebras, which builds hardware specifically designed for extremely fast AI inference. That partnership is what makes these numbers possible.

Competitors have been working the same angle. Anthropic already offers a fast mode for Claude, though it does not reach the speeds OpenAI is claiming here. As models converge on similar capability levels, speed and cost are becoming the differentiators companies actually compete on.

Ultrafast is in preview right now and only available to a small group of customers. OpenAI says access will expand as capacity grows.

Lovable has confirmed a new $400 million Series C round at a $13.3 billion valuation, led by Menlo Ventures and the Scaleup Europe Fund with more than a dozen other investors taking part. To put the pace in perspective, the company's previous round closed in December at a $6.6 billion valuation. It has doubled in eight months.

The revenue growth explains the jump. Lovable hit $500 million in annualised run rate revenue in June, up from the $400 million it reported earlier this year. The usage numbers are just as striking. The platform now hosts 60 million projects that pull in 900 million monthly visitors, which is the kind of traffic most established software companies never reach.

What is interesting about how Lovable has scaled is that it has not just been riding on other companies' models. The company now offers its own in-house trained AI model alongside the usual frontier options from Anthropic and Google, and has been publicly arguing that letting users pick from a menu of models is a dead end. In June it also signed a multiyear deal with Google Cloud that increased its usage fivefold, giving it the infrastructure to support that growth.

Lovable has also started acting like an investor itself, backing other European startups including Danish company Atech, which is building vibe-coding software for designing hardware.

For a company that was barely a name eighteen months ago, going from near zero to $500 million in revenue and a $13.3 billion valuation is one of the fastest arcs in recent software history. Whether the vibe coding category holds up long term is still an open question, but Lovable is currently the clearest evidence that it is more than a passing trend.

Sundar Pichai announced on X that the Gemini app has passed 1 billion monthly active users, making it one of Google's fastest-growing products ever and the 14th Google product to reach the billion user mark. That last detail is worth sitting with. Very few companies have one product at that scale. Google now has fourteen.

The number puts Gemini level with ChatGPT, which hit 1 billion monthly active users back in June. Considering Gemini was widely seen as playing catch-up not that long ago, closing that gap in a matter of months is a serious turnaround.

The 1 billion refers specifically to the Gemini app and does not include people using Google's AI through other channels. AI Mode in Search separately has over 1 billion monthly active users of its own, so the total number of people touching Google AI in some form is considerably higher.

The usage details Google shared are the more interesting part. 63% of Gemini users talk directly to the assistant using the voice feature, which is a striking number and suggests voice is becoming the default way people interact with AI rather than a novelty. Gemini also now generates more than 150 million images every day. And notably, growth is not confined to Google's own ecosystem, with more than 100 million active users on iOS.

Google reported over 950 million monthly Gemini users at its Q2 2026 earnings call, with daily active users tripling over the past year. The billion milestone landed just before its Made by Google event, where more Gemini features are rolling out across Pixel devices. Google's strategy of embedding Gemini everywhere, across Search, Workspace, Android, and now cars and laptops, is clearly working.

Create short films, marketing videos, and ads from scripts, references, or product URLs. Topview's AI Video Agent plans and creates every scene on one