Changing your home’s address on Apple Maps can be almost as stressful as moving home in real life, but there’s one sure-fire way to get the job done without pulling your hair out.

The Apple Maps app is a great way to get around, whether you’re using it while driving via CarPlay or just walking. It also makes it easy to quickly get directions to the places that you visit often, like your home or where you work.

But if you’ve moved home and ever tried to change your address in your iPhone‘s Maps app, you’ll know it can be tricky. The job can be complicated considerably if you also have your home address pinned, too.

Thankfully, changing your home’s address is much easier when you keep in mind two main things. The first is that your home’s address isn’t normally set by the Maps app itself. And the second is that pinned addresses in the Maps app don’t automatically update, nor can they be edited.

It’s easy to see where the confusion comes from, but don’t worry. We’re going to clear everything up right here.

Your ‘Home’ address has a home of its own

The first thing to remember is that your iPhone uses your contact card as the source of your home address whenever it is needed. It’s this address that is used when you make purchases online, fill out online forms, and for anything else that requires an address.

Remember to update your pinned location in Apple Maps after changing your home address.

If you move home, it’s this address that you need to change. And thankfully, it’s easy to do.

1. Open the Contacts app on your iPhone and tap My Card at the top of the screen.
2. Tap Edit in the top-right corner of the screen and then scroll down to the address portion of your contact card.
3. Tap the address fields and enter your new details.
4. Tap the checkmark in the top-right corner of the screen to save your new address.

That’s all there is to it, your iPhone now has your new address. However, if you have your home set up as a pinned location in Maps, you’ll also need to go a step further.

Unfortunately, Maps won’t automatically update the pin to reflect your new address. Pinned addresses also can’t be edited, so you’ll have to delete the pin and add a new pin all over again.

I recently got to watch what happens when you jailbreak some of the world’s most powerful artificial intelligence models.

Don’t worry—this AI manipulation wasn’t used to hack anyone or build a nuclear bomb. I simply got to see firsthand how vulnerable some frontier models are to ditching their safety guardrails.

FAR.AI, an AI safety nonprofit based in California, built a tool that takes a range of problematic prompts and generates more than a thousand different versions in an attempt to identify functioning jailbreaks. I saw some models generate a detailed plan for launching a cyberattack on an imaginary hydroelectric dam, among other things. Often, it involved trying dozens of prompts, with models rejecting many of them out of hand.

I chatted with FAR.AI in advance of a new report, which saw the group test the safety guardrails of models from four popular US companies: Anthropic’s Claude Opus 4.8 and Fable 5; OpenAI’s GPT 5.5 and 5.6; Google’s Gemini 3.1 Pro; and Grok 4.3 and 4.5, from Elon Musk’s newly combined SpaceXAI. It auto-generated prompts designed to trick the models into doing potentially harmful things, like generating software exploits and providing details for developing chemical or biological weapons.

The report found that Grok was most vulnerable to jailbreaks, with 448 jailbreaks found, followed by Gemini, with 249 found, while Claude, Fable, and GPT were impervious to the attacks. However, that doesn’t mean those models are immune to more sophisticated jailbreaks, which may involve interacting with a model in more complex ways, according to FAR.AI and other experts.

The report also calculated the cost of getting models to misbehave by using another AI model to automatically generate different jailbreaks. The results are dirt cheap, all things considered—$58 to jailbreak Grok and $278 to jailbreak Gemini.

“AI models right now are less regulated than restaurants,” says Adam Gleave, the CEO of FAR.AI and an expert on AI safety and alignment.

Gleave says that the findings demonstrate the need for externally imposed standards and regulations. “Talk of relying on voluntary commitments, that AI companies are going to be able to self-regulate, is nonsense,” he says.

But Gleave also believes that the findings show that models can be systematically tested for safety. “There’s an optimistic angle here,” he says. “Defense and safety really are possible.”

Rohin Shah, the director of AGI safety and alignment at Google DeepMind, says the results of the report “should not be interpreted as a comprehensive assessment of Gemini’s safety and security,” because not all jailbreaks are equally severe.

“We are constantly working to improve our safeguards,” Shah says. “We conduct extensive red teaming and evaluations across severe misuse risks and apply multiple layers of protection throughout development and deployment.”

“These findings reflect the sustained investment we’ve made in our safeguards,” Anthropic spokesperson Michael Aciman tells WIRED. “We continue to evolve our safety systems as these attacks become more sophisticated.”

“Jailbreaks are an ongoing challenge across the industry, and we continuously strengthen our safeguards as attack techniques evolve. We rigorously test our models against new threats and use those findings to improve our protections,” OpenAI spokesperson Gaby Raila said in a statement to WIRED.

SpaceXAI did not respond to WIRED’s request for comment.

Recently passed state laws in California and New York require frontier AI developers to publish safety reports, and soon, an Illinois law will require those companies to have their safety practices evaluated by third-party auditors. But the federal government hasn’t yet passed any specific safety requirements, and chaos has ensued as the industry—and officials—try to figure it out.

Microsoft is in a unique position as AI overtakes the tech industry. It’s one of the world’s largest cloud providers and software-as-a-service companies, while also holding valuable stakes in the two biggest AI labs, OpenAI and Anthropic.

Those incentives are starting to clash as Microsoft posts blockbuster financial results. The company just reported an extremely profitable quarter with $90 billion in revenue and net income of $35.8 billion. For the fiscal year, which ended June 30, Microsoft reported $331.8 billion in revenue with a net income of $133.7 billion for the year.

And CEO Satya Nadella is not about to let the trajectory of Anthropic and OpenAI — which are expanding into applications and agentic infrastructure that could ultimately let them own customer relationships — derail that kind of cash.

Nadella has been preaching to enterprises to use multiple models and to stop relying on the frontier AI labs for the agentic harness/app layer.

Doing so is dangerous, he’s been saying, because it requires companies to share too many of their internal secrets with model makers of dubious trustworthiness. He knows his customers. Enterprise IT fears both data leaks and being locked into a vendor.

Now he has openly told Wall Street analysts during the company’s quarterly conference call Wednesday that this is an opportunity for Microsoft to sell customers its own homegrown models, alongside agents, AI security and more, while promising lower costs.

In other words, he’s pitching Microsoft as an alternative to many of the upscale services that OpenAI and Anthropic are developing for their own growth.

When UBS analyst Karl Keirstead specifically asked Nadella to weigh in on the open vs. closed-sourced debate roiling the AI industry, and how Microsoft will benefit from it, Nadella came out swinging.

“The goal is to have the firm be in control of their own destiny,” the CEO said of enterprises. “We are very, very clear about the architectural sort of design of the platform, which is you got to keep your harness separate from the model … that means any model at any given time is swappable.”

Microsoft, of course, sells a menu of harnesses (aka AI agents), too, under the Copilot name, including its coding agent GitHub Copilot. Coding agents are where much of the AI dollars are being spent today.

And he used the high-profile incident from last week as proof of his warnings.

“If you look even at the Hugging Face incident, the biggest thing that we should take away from that is you can’t sort of depend on any one model,” Nadella said. “You will maybe need multiple models to even remediate some challenges that get caused by one model. Like that’s the way to think about it, right? Which is you can’t be subject to a refusal of one model.”

The incident involved an unreleased model from OpenAI breaking out of its sandbox and successfully mounting a full-scale hack on Hugging Face, all in pursuit of besting a benchmark. Trying to understand what happened, Hugging Face at first tried to use a private frontier model (which it hasn’t named) that refused to help it. So it turned to the Chinese open-source model Z.ai GLM 5.2 to analyze logs and defend its infrastructure. The incident has so shocked the industry that even Sam Altman is now saying that maybe AI development should slow down a bit.

Nadella also made clear that Microsoft is happily selling its own homegrown models, the MAI family, on its own homegrown AI chips, Maya, and pitching them as cheaper alternatives.

“Every customer wants the right model for each task based on quality, latency, cost, and compliance. We offer the broadest model catalog in the cloud with over 11,000 models, including the leads from OpenAI, Anthropic, Mistral, xAI, as well as our own MAI family,” he said.

He added: “We’re also accelerating our own model development. We announced more than a dozen new models across image, voice, transcription, coding, security, including our first reasoning model, MAI thinking one, all with cost-efficient inference at the core for the enterprise use cases. We are co-designing these models with our silicon, and we are seeing 40% better performance per watt when running MAI models on Maya 200.”

As for Mythos? Nadella pointed to Microsoft’s new Mythos competitor announced earlier this week, MAI Cyber One Flash. It “achieves better performance than the much larger Mythos model, but at half the cost when combined with our multi-agent security harness,” he said.

Sure, the Microsoft CEO says that enterprises should use the frontier models that OpenAI and Anthropic offer in their mix. But his bigger message is: don’t trust them enough to rely on them.

Wild Bio, founded in 2021 by Dr Ross Hendron and Irish man Prof Steve Kelly, specialises in crop genetics by researching genetic solutions that developed through evolution.

University of Oxford spin-out Wild Bioscience, which develops improved crop varieties using AI and precision breeding, has acquired wheat breeding business F1 Seed for an undisclosed amount.

By acquiring F1 Seed, Wild Bio aims to become a “fully integrated, end-to-end UK seed company”, forming a single business able to design, breed and deliver new wheat varieties to growers.

Wild Bio, founded in 2021 by Dr Ross Hendron and Irish man Prof Steve Kelly, specialises in crop genetics by researching genetic solutions that developed through evolution to improve crops in the past, and then using these discoveries to design simple fixes for modern crops.

The start-up does this by building unique plant biology datasets and applying AI and machine learning tools to find “signatures” for how evolution has improved plants. The Wild Bio platform can then target specific “nature-tested” traits and activate them in crops such as wheat.

With this acquisition, Wild Bio will be able to utilise F1 Seed’s wheat breeding capabilities and access its diverse germplasm – genetic resources such as seeds, tissues and DNA sequences – pipeline, which has been developed for more than a decade.

Wild Bio’s goal is to deliver conventionally bred, optimised wheat varieties directly to UK farmers by 2027, followed by the first precision-bred lines in field trials by 2028.

F1 Seed already has six wheat varieties on the UK market, primarily focused on animal feed and distilling, and plans to launch three more later this year. Of the 82 new wheat varieties entering official UK national trials this year, 12pc – 10 entries – come directly from F1 Seed’s pipeline, according to a press release on the acquisition.

“This is the most exciting news for wheat growers since Cambridge University established the Plant Breeding Institute in 1912,” said Bill Angus, founder and director of F1 Seed. “That organisation established the blueprint for integrating new traits into cereals.

“Now over a century later, armed with an array of novel traits, we are well placed to move wheat breeding into a new and exciting era. Deploying traits has been the lifeblood of wheat breeding – now we have the opportunity to accelerate this in UK germplasm.”

Wild Bio said the acquisition comes as the UK’s farming industry hits a tipping point, as wheat yields have remained flat for three decades, while input costs have risen and the climate crisis has introduced “unprecedented volatility”.

In fact, a report from last November found that one-third of surveyed UK farmers made no profit in the previous year, with more than half considering leaving the industry.

With the record-breaking heatwaves and drought seen in recent years affecting wheat yields significantly, Wild Bio said it will be able to help UK farmers by providing seeds designed specifically for the country’s conditions.

“Wheat is Britain’s biggest crop and feeds a fifth of humanity. But on a changing planet, growing it is only getting harder,” said Hendron, who is also Wild Bio’s CEO. “That’s the challenge we exist to meet.

“As a new independent seed business that brings breeding and gene editing together, we can focus on the traits that growers genuinely need and design towards them with precision. Bill and his team have built something remarkable; pairing their strengths with our evolutionary platform gives us the full stack of capabilities we need to develop the varieties that will bend the yield curve upwards again.”

Last October, Wild Bio raised $60m in a Series A round led by the Ellison Institute of Technology, founded by Oracle’s Larry Ellison. The company’s acquisition of F1 Seed was largely financed by this funding, according to Wild Bio.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Artificial intelligence is rapidly becoming part of everyday infrastructure–in some places. It helps write emails and software code, filters job applications, powers recommendation systems, and is increasingly being integrated into education, health care, finance, and public administration. Industry leaders talk about “AI for everyone,” while governments rush to publish national AI strategies and build sovereign compute.

Yet over the past decade, working on digital inclusion and digital literacy projects in regions from Europe to sub-Saharan Africa and Southeast Asia, I’ve seen the same pattern repeat: Each new wave of “transformative” technology lands on a landscape already stratified by connectivity, skills, and institutional capacity. The current AI wave is no exception. If anything, it amplifies those underlying fractures.

Still, some countries are exploring ways of participating in AI development without directly replicating the frontier-model race dominated by the United States and China. Recent developments in South Africa and Indonesia illustrate both the possibilities and challenges. The stakes extend far beyond access to AI. Countries that remain primarily consumers rather than creators of AI risk losing opportunities to build local innovation ecosystems, strengthen public-sector capacity, and ensure that their own languages, cultures, and societal priorities are reflected in AI systems. In this sense, the AI divide is also becoming a divide in economic opportunity and technological influence.

AI compute is clustering in a few places

Recent analyses from Stanford University’s 2026 AI Index report that the United States alone hosts more than 5,000 data centers, over 10 times as many as any other single country. Because AI workloads are increasingly performed on cloud platforms rather than local infrastructure, this concentration of compute also becomes a concentration of dependency. According to World Bank data, in 2023 the United States accounted for roughly 87 percent of global exports of cloud computing and data-storage services.

For most countries, this means that AI development is not just technologically but commercially and geopolitically outsourced and out of their control. The result is an AI ecosystem where a small number of states and firms host the computational engines that power globally deployed systems.

Systems trained, standardized, and governed within a narrow set of institutional and linguistic environments may struggle to serve a genuinely global public.

Skills and AI literacy are deeply stratified

Even where connectivity and cloud access exist, not everyone is equally positioned to make use of them. Across the Organisation for Economic Co-operation and Development (OECD) countries, only around 40 percent of adults possess more than basic digital problem-solving skills, while advanced computational and AI-related competences remain concentrated among highly educated workers and technology-intensive sectors.

At the same time, governments are racing to integrate AI into education, often starting at higher levels of schooling. UNESCO has reported growing efforts worldwide to integrate AI into education, while support for AI literacy in primary and lower secondary education, as well as ethical training for educators, remains uneven.

Those with robust schooling, advanced digital skills, and stable connectivity are best positioned to treat AI as a tool to extend their capabilities. Recent OECD survey data show that participation in AI-related training remains strongly stratified by educational attainment: 36 percent of respondents with tertiary education reported undertaking AI-related training in the previous year, compared with just 18 percent of those with upper-secondary education. Those on the wrong side of the divide are more likely to experience AI as an opaque system acting upon them, from algorithmic welfare systems such as the Dutch childcare benefits scandal to AI-assisted hiring tools such as Amazon’s discontinued AI recruiting system, rather than as a technology they can actively interrogate or shape.

Investment and governance: Who gets a seat at the table?

The core agenda-setting power often remains with a narrow set of industry actors and a small group of technologically advanced states. Most other countries remain in a perpetual catch-up posture, adapting imported models, standards, and templates for “trustworthy AI” to their own contexts, and may have limited local capacity to assess trade-offs or propose alternatives.

In countries such as Indonesia and South Africa, communities generate data at massive scale yet still have little voice in how AI systems are designed, governed, or deployed. Their languages are underrepresented in training data; their institutions are under-resourced in regulatory forums; their experiences rarely feature in benchmark datasets. For many countries in the global South, participation in AI still occurs largely through adapting imported systems rather than shaping how those systems are designed, governed, or deployed.

In South Africa, the Department of Communications and Digital Technologies released a draft national AI policy in April 2026, proposing new oversight institutions. The department withdrew the draft days later after a journalist discovered that at least six of its academic citations did not exist, apparently AI-generated hallucinations. The minister called it “an unacceptable lapse.“ The episode sharply illustrates the gap between AI governance ambition and the institutional capacity needed to implement it, though the new AI panel the country has since constituted has a chance to use South Africa’s unique leverage.

Indonesia presents a case of deliberate, if constrained, public-sector agency. The National Research and Innovation Agency (BRIN) which now leads AI implementation under the national strategy, has built practical AI tools aimed at underserved communities rather than frontier capabilities, including an app that uses satellite data and machine learning to help artisanal fishermen locate schools of fish, multilingual language models trained on Indonesian and local languages such as Javanese and Sundanese, and AI chatbots deployed in government services. In August 2025, the Ministry of Communication and Digital Affairs released a national AI road map with a target of training 100,000 AI-skilled workers annually.

The choice is not simply between “AI superpower” and “passive recipient.”

Regional cooperation may also become increasingly important. In 2024 African ministers adopted a Continental AI Strategy and African Digital Compact. Participants in the April 2025 Global AI Summit on Africa in Kigali explored how regional coordination, local-language AI models, public universities, and open-source ecosystems might reduce long-term dependence on externally developed AI systems.

A different way to think about the AI divide

None of this means that people should slow or abandon AI, nor that cloud concentration or venture capital are inherently bad. Instead, when we talk about an “AI revolution,” we should also ask who can shape it and who can merely adapt to it.

Digital-divide debates once focused on devices and connectivity, later expanding toward skills and outcomes. But the current AI wave adds another layer: disparities in who can meaningfully participate in deciding what AI is for, which problems it is meant to solve, and which social priorities it ultimately serves.

For engineers and policymakers, this raises difficult but necessary questions. Are they designing AI systems and infrastructures that broaden, rather than narrow, participation in shaping technological change? When governments roll out national AI strategies or integrate AI into public services, whose constraints, languages, and institutional realities are they including?

Many observers frame the current AI moment as a competition. But technological competition is never only about speed. It is also about who can influence the direction of change.

AI is already spreading globally. The deeper question is whether the technologists and policymakers responsible for it will ensure that meaningful participation in shaping that future will spread as well.

• Verizon signs $1 billion dark fiber data center connectivity deal with Google
• The telecoms company plans to use its fiber assets to connect AI data centers across the US
• Fiber network growth is replacing copper lines and activating previously unused dark fiber lines

Verizon has completed a $1 billion deal with Google to provide data center connectivity using pre-existing dark fiber infrastructure. The move appears to be the first of several arrangements to leverage dark fiber and the desire for AI data centers to collaborate on compute tasks.

CEO Dan Schulman highlighted how copper is being ripped out of central offices to replace the outdated cabling with modern fiber to meet AI demand in workplaces. Verizon previously announced its AI Connect initiative, which aims to support AI infrastructure growth, and the Google deal appears to be the first step in that initiative.

Shark’s PowerPro has just hit an all-time low price, that matches this discount we saw during Prime Day.

The drop takes the Shark PowerPro cordless vacuum down to £169.99, a full £80 off its usual £249.99 price tag.

This works out at 32% saved on a machine built to handle carpets, hard floors, upholstery and even the inside of a car.