AWS Is Building An Internet For The Machines

Cloud infrastructure has long been designed around humans who search, click, scroll, and stream in a steady and predictable fashion. AI agents behave differently. They can unleash a swell of activity, spinning up multiple sub-agents that query hundreds of databases, search documents, and call APIs in seconds and then disappear as quickly as they arrived.

Under that premise, Amazon is redesigning a core piece of its cloud infrastructure.

On 28 May, AWS launched its next generation of OpenSearch Serverless, a fully managed search and vector database — essentially a system for storing and retrieving information at scale — that’s designed specifically for agentic workloads. AWS says the new system can instantly scale up when agents trigger tasks and scale back down to zero when idle.

The launch reflects a growing realization across the tech industry: Infrastructure originally designed for a human-driven internet doesn’t work as well in a world increasingly populated by agents.

While AI agents still represent a relatively small portion of internet activity, machine-generated traffic is already significant, and poised to grow. Cloudflare says bots accounted for 31 percent of overall HTTP traffic over the last six months. AI crawlers, search engines, and assistants made up roughly a quarter of all bot requests during that period.

"Non-human traffic will exceed human traffic sometime in the first half of 2027," said Lai Yi Ohlsen, senior product manager at Cloudflare, to TechCrunch.

At Google’s I/O developer conference last week, the company said users will be able to start delegating tasks to AI systems, like researching purchases, booking travel, browsing the web, and interacting with apps. But the buck doesn’t stop at consumer-focused AI agents. Enterprises are increasingly deploying agents internally and for their customers, creating new kinds of machine-generated traffic behind the scenes.

As a result, cloud providers and infrastructure companies have been reckoning with how to adapt systems built for humans to a world of agents that are constantly and autonomously retrieving information, invoking tools, and generating machine-to-machine traffic.

That’s where AWS’s new OpenSearch Serverless comes in.

"The timing is straightforward. Agents are moving from experimentation into production, and they create traffic patterns that previous infrastructure simply wasn’t designed for," Tia White, general manager for Amazon OpenSearch Service, told TechCrunch. "They spike without warning, they go idle without notice, and enterprise needs search that keeps up without paying for empty or idle compute."

The key technical change with this new generation is that it decouples compute from storage, allowing compute to scale up in seconds to accommodate agent traffic bursts and to scale down to zero, so customers pay US$ 0 when agents are idle.

"Previously, even in our prior Serverless version, you had to have at least one instance operational and running because storage and compute were coupled," White said. "You couldn’t just automatically spin up [compute] at the rate you needed to, so you always had idle compute reserved for your workload, whether you were using it or not."

Think of it like always paying for a parking space, even when you’re not using it. With AWS’s upgraded Serverless, it’s more like paying for a metered parking spot.

At launch, OpenSearch Serverless will integrate natively with AI development platforms like Vercel and Kiro, so developers can deploy production-ready search and vector backends for agents without managing infrastructure.

The shift is emerging across the cloud industry. Databricks and Snowflake are repositioning themselves as AI memory and retrieval systems for enterprise data. Microsoft has rolled out updates to Azure designed to handle AI agent bursts and share memory between agents. Cloudflare, in a similar vein to Amazon, last month introduced infrastructure aimed at giving agents persistent environments and instant scalability.

Read More

Oura Ring 5 Offers More Health Tracking Features

Two years after the Oura Ring 4, Oura is back with a new smart ring that aims to make health tracking smaller, smarter, and more proactive. The newly announced Oura Ring 5 is now available for preorder and begins shipping 4 June. It features a slimmer design, stronger durability, upgraded sensors, and a growing list of AI-powered wellness tools.

This model joins the growing trend of screen-free fitness trackers and minimalist wearables that are becoming increasingly popular among users who want health insights without smartwatch distractions. Starting at US$ 399, the Oura Ring 5 builds on Oura’s wellness-first approach while introducing new features focused on fitness, sleep, and long-term health monitoring.

Oura is introducing a new feature called Health Radar. It’s a system that continuously monitors your biometric patterns and looks for subtle changes that could signal developing issues.

Health Radar currently has two sub-features: Blood Pressure Signals and Nighttime Breathing monitoring.

Blood Pressure Signals looks at trends that may be linked to cardiovascular strain, while Nighttime Blood Pressure tracking monitors how blood pressure changes during sleep. Additionally, Nighttime Breathing provides a 30-day view of breathing disturbances that may affect your sleep quality.

The company is also expanding support for metabolic health. The new GLP-1 medication tracking tools will let you log doses, side effects, and weight changes. Plus, the Lab Uploads will allow subscribers to import blood test results and compare health markers over time.

Oura is also expanding its AI-powered Advisor platform, which serves as a wellness assistant and offers more personalized health guidance. You’ll be able to ask health questions, receive tailored recommendations, and, through a new partnership with Counsel Health, connect with licensed physicians from within the app.

A new live activity mode will allow you to track heart rate, pace, and distance in real time through phone lock-screen widgets. And when you engage in workouts where wearing a ring may not be practical, such as strength training or powerlifting, Oura can now collect your metrics via third-party heart rate monitors.

The brand has also improved the Automatic Activity Detection feature, making it easier for the ring to detect low-motion workouts like Pilates.

According to Oura, the Ring 5 uses LEDs that are four times as powerful as those in the previous model. The redesigned sensors sit closer to the skin and use 12 independent signal pathways to improve accuracy across different finger shapes and skin tones.

Additionally, the new ring has a better battery, and you can expect between six and nine days of battery life on a single charge, depending on usage. The brand is also launching a new portable charging case, which will cost US$ 99. It can provide roughly five additional charges, allowing you to enjoy about a month of extra battery life.

The Oura Ring 5 is 40 percent smaller than the Ring 4, measuring just 6.09 mm wide and 2.28 mm thick. The company is calling it the world’s smallest smart ring, designed to feel lighter and more comfortable for around-the-clock wear.

Read More

Google Plans To Release Millions Of Mosquitoes

California could soon become a testing ground for one of Google’s most ambitious public health projects yet.

The tech giant is seeking federal approval to release up to 32 million specially treated mosquitoes in California and Florida over the next two years as part of an effort to reduce the spread of mosquito-borne diseases, including West Nile virus, St. Louis encephalitis, dengue, Zika, chikungunya and yellow fever.

The proposal is currently under review by the US Environmental Protection Agency, which is accepting public comments through 5 June before deciding whether to issue an experimental use permit.

Regulators have not announced where any mosquito releases would occur if the plan is approved.

Researchers say the latest proposal targets Culex mosquitoes, a species known for transmitting West Nile virus and St. Louis encephalitis.

According to the Centers for Disease Control and Prevention, West Nile virus remains the leading mosquito-borne disease in the U.S.

Those viruses are already established in California, where they circulate naturally among local bird and mosquito populations.

On Friday, a positive sample of West Nile virus was confirmed in Riverside County.

The project is part of Google’s little-known Debug initiative, launched more than a decade ago to develop new technologies aimed at reducing populations of disease-carrying mosquitoes.

Rather than releasing biting insects, the company plans to release male mosquitoes infected with Wolbachia, a naturally occurring bacteria. When the infected males mate with wild female mosquitoes, the offspring do not survive, helping suppress mosquito populations over time.

Because only female mosquitoes bite humans, experts say the releases would not increase the number of biting mosquitoes.

Read More

UK's Rudest Chalk Figure Gets A Makeover

For centuries, the Cerne Abbas Giant has been hard to miss because of its rude features.

The 55-metre chalk figure, cut into a hillside near the village of Cerne Abbas in Dorset, shows a naked, club-wielding man whose outline has made him one of the UK's most instantly recognizable historic landmarks.

But the National Trust, which owns and manages the site, says changing weather patterns are making it harder to keep the Giant prominent on the hillside.

National Trust staff and volunteers will this week pack tonnes of new chalk onto the figure to restore the crisp whiteness of his outline.

Luke Dawson, a National Trust ranger who helps look after the site, says heavier winter rains are washing chalk from the slope more quickly, while mild, damp conditions give algae more chance to grow.

He says this wetter weather has been having "a dulling effect" on the Giant's outline, leaving it greener and less distinct between maintenance work.

The Trust is cautious about attributing the changes directly to climate change at a single site.

"It's one of these things we cannot really prove," says Dawson. "It is more just observation of what we are seeing up there."

The charity has cared for the Giant since 1920. Its rangers and volunteers keep the outline defined by rechalking the figure every decade or so to protect it from weeds and erosion. And between chalking it uses sheep to keep the grass short.

But the Trust says that coupled with heavier winter rains, the frequent dry spells in summer, mean the grass grows back more slowly and can leave the chalk edges more exposed and vulnerable to erosion.

The rechalking could take up to 15 days to complete. Around 300 National Trust staff and volunteers will be involved, carrying about 17 tonnes of fresh chalk up the steep hillside, which in places has a gradient of roughly one in three.

The work is physically demanding, especially in the exceptional heat the UK has experienced in recent days. The old chalk is carefully dug out before fresh material is packed into the Giant's outline by hand - a process the Trust says has changed little for generations.

The Giant's naked, club-wielding form has fuelled centuries of speculation - "a real ding-dong" according to local historian Ian Denness. Some argued he was an ancient fertility figure, others a Roman Hercules, or even a later satire of Oliver Cromwell.

But scientific analysis of sediments published by the National Trust in 2021 suggested the figure was probably first cut in the late Saxon period, between around 700 and 1100AD - much later than the prehistoric or Roman origins once imagined.

However, this finding has not settled the question of his significance.

Read More

Google's AI Still Can't Spell, Including The Word "Google"

And though the Gemini-powered technology has improved its accuracy dramatically over the last two years (unfortunately for publishers), AI Overviews still gets basic questions wrong. And that includes spelling tests.

Google's AI tools remain abysmal at answering questions about spelling, having gone viral two years ago for responding to the question "how many r's are in the word strawberry?" incorrectly. But it's still bad. Last 26 May, X user Naomi Rohatyn tested the large language model's (LLM) current ability to answer to a spelling question.

"There are exactly 2 'e's in the word "astronomical" (a-s-t-r-e-n-o-m-i-c-a-e-l)," replied AI Overview.

How many Ps are in Google? According to Google, there are two.

There’s also is also "exactly 1 'r' in the word 'poop'," Google’s AI Overview says, as well as two ‘d’s in the word journalism, yet spelled it: j-o-u-r-n-a-d-i-s-m. Google did at least identify that there is one P in the last name of the U.S. president, but spelled it as t-r-p-u-m.

Considering users are less likely to click on links when an AI summary appears in the results, surely the information provided in AI Overviews should be accurate. But it's complicated.

AI chatbots need exact context and specifics to answer as well as they can, so surely spelling words within their training data seems easy. However, things get knotty when you ask an LLM to consider words letter-by-letter, as the model will process text in chunks rather than individual characters (it's called tokenisation).

"Counting within words has been a known challenge for LLMs, and we’re working to fix this particular issue," Google told TechCrunch in an emailed statement.

These basic spelling errors may seem familiar. LLMs, the kind of artificial intelligence that powers chatbots and other text-generators, are not built to understand spelling. It’s been a running joke for years that whenever a company unveils a new AI model, you should ask it how many 'r's are in the word strawberry. These AI models — which can code an app in seconds, or solve problems that have stumped mathematicians for decades — are about as good as a kindergartener at spelling.

Google’s AI overview woes reach beyond silly spelling mistakes though. Google already patched an issue from last week in which searching the word "disregard" would yield what looked like a dictionary definition of the word, only the definition was shown as, "Understood. Let me know whenever you have a new prompt or question!" But these spelling errors have remained amusing because they’re so difficult to quash.

"LLMs are based on this transformer architecture, which notably is not actually reading text. What happens when you input a prompt is that it's translated into an encoding," Matthew Guzdial, an AI researcher and assistant professor at the University of Alberta, told TechCrunch. "When it sees the word ‘the,’ it has this one encoding of what ‘the’ means, but it does not know about 'T,' 'H,' 'E.'"

The token-based architecture that powers LLMs like Google’s AI overview is inherently limiting, and researchers haven’t been optimistic that they can solve the spelling problem.

Read More