A series of incidents involving Waymo’s autonomous vehicles has highlighted what happens when driverless systems fail in complex real world situations and how much they still rely on human intervention to recover.

A Technology Built For The Road Meets The Unexpected

Waymo’s robotaxi service has expanded rapidly across multiple US cities, now delivering hundreds of thousands of paid rides each week. The company positions its system as a fully autonomous driving service, designed to operate without a human driver behind the wheel.

However, recent incidents show that when situations fall outside expected conditions, vehicles can struggle to respond. In several reported cases, Waymo vehicles have stopped, hesitated or behaved unpredictably during emergencies, requiring intervention from police officers or other first responders.

One widely reported example from August 2025 involved a highway fire in California, where traffic was redirected in an unusual way. A Waymo vehicle was unable to adapt to the change, eventually stopping and requiring a police officer to manually move it out of the way.

When Autonomous Vehicles Cannot Proceed

The most significant issue here seems to be what happens when the system cannot decide what to do next.

Autonomous vehicles are designed to prioritise safety, which often means stopping when uncertainty is too high. While this reduces the risk of collisions, it can create new problems, particularly in fast-moving or emergency situations where standing still is not a viable option.

In multiple incidents, it seems that autonomous vehicles have effectively become obstacles in live environments, blocking traffic or delaying access for emergency services until human intervention takes place.

Human Support As The Fallback

To manage these situations, Waymo relies on human support systems behind the scenes.

The company uses Remote Assistance teams who provide contextual guidance when the vehicle encounters something it cannot resolve. According to Waymo, these workers do not drive the vehicle. Instead, they support decision-making. As the company explains, Remote Assistance agents “provide advice and support to the [vehicle] but do not directly control, steer, or drive the vehicle.”

This model is designed to ensure that the automated system remains in control at all times. However, it also means that when the system reaches its limits, recovery can depend on how effectively this human support is integrated.

Where Things Can Go Wrong

Even with this support in place, errors can still occur. For example, in one case under investigation in Austin, Texas, in January this year, a Waymo vehicle approached a stopped school bus with warning lights active. The system requested input from a remote assistant, who it is alleged incorrectly confirmed it was safe to proceed. The vehicle then moved past the bus while children were boarding, an action that would normally be illegal for a human driver.

Other reported incidents show a different type of failure, where no safe path is identified at all. In these cases, vehicles have remained stationary until physically moved, sometimes by police or other first responders.

All this has led to local officials raising concerns that this places an unexpected burden on public services. For example, in San Francisco, emergency management leaders warned that responders were becoming a default support function for autonomous vehicles, something they described as unsustainable.

Scaling The Problem Alongside The Technology

It seems that these challenges are becoming more visible as Waymo scales its operations.

The company operates thousands of vehicles and is expanding into new cities, increasing the number of unpredictable environments its systems must handle. It has said that around 70 Remote Assistance agents support a fleet delivering more than 400,000 rides per week.

In its response to US lawmakers, Waymo reiterated that Remote Assistance is limited in scope, stating that agents “provide advice only when requested by the automated driving system on an event-driven basis” and do not take control of the vehicle.

As deployment grows, the question is not whether incidents will occur, but how frequently and how effectively they can be resolved without external intervention.

Balancing Autonomy With Accountability

Waymo maintains that its system is designed to prioritise safety, even if that means stopping when conditions are unclear. The vehicle can also ignore human input if it conflicts with its own assessment, reinforcing that it remains the primary decision maker.

The company also states that “Waymo’s service does not rely on remote drivers,” emphasising that human involvement is limited and controlled.

However, the pattern of real world incidents seems to suggest that full autonomy still depends on multiple layers of human support. When those layers are not sufficient, responsibility can extend beyond the company itself to public infrastructure and emergency services.

What Does This Mean For Your Business?

For UK businesses, this highlights a critical aspect of automation that is often overlooked, namely what happens when systems fail or reach their limits.

Autonomous technologies are not just defined by how they perform under normal conditions, but by how they behave when they cannot proceed. Stopping safely is one outcome, but in operational environments, recovery is just as important.

It seems that human oversight, fallback processes and clear responsibility models remain essential. Businesses adopting automation will, therefore, need to plan not only for success scenarios, but also for failure scenarios, including how issues are resolved quickly and safely.

There is also a wider accountability question here. When automated systems interact with public environments, any gaps in ownership can become visible very quickly.

The Waymo case shows that the real test of autonomous systems is not when everything works, but how they respond when it doesn’t.

A new chemistry breakthrough could allow people to use long fingernails on touchscreens, addressing a long-standing usability issue with modern devices.

Why Fingernails Don’t Work On Touchscreens

Most modern smartphones and tablets use capacitive touchscreens, which rely on tiny electrical fields across the surface of the display. When a conductive object, such as a fingertip, disrupts that field, the device registers a touch.

Fingernails, however, are not conductive. This means taps made with the nail itself are not recognised, forcing users to adjust how they interact with devices. For people with long nails, this often results in awkward movements or reduced accuracy.

The issue is actually more widespread than it first appears. It also affects individuals with heavily calloused skin, where reduced conductivity can lead to unreliable touch response.

A Chemistry Led Solution

The new approach has been developed by a student researcher working with a supervisor at Centenary College of Louisiana and presented at a meeting of the American Chemical Society.

The idea is simple in principle, i.e., to create a nail coating that allows fingernails to interact with a touchscreen in the same way as skin.

As part of the research, the team experimented with more than 50 additives across multiple nail polish formulations. Their goal was to find a combination that could introduce just enough electrical interaction to register a touch, without compromising safety or appearance.

The motivation for the work came from a real-world need. As the researchers noted, when they explored the problem, the response was immediate: “would a touchscreen-compatible nail be useful?” The answer, they said, was “a resounding ‘yes, please!’”

How The Nail Polish Actually Works

Rather than making the nail directly conductive in the traditional sense, the formulation works through a different mechanism.

The researchers identified two key ingredients, taurine, commonly found in dietary supplements, and ethanolamine, a simple organic compound. When combined in a specific way, these ingredients enable a small movement of electrical charge across the nail surface.

This is enough to create a change in capacitance, allowing the touchscreen to detect contact.

According to the researchers, “our final, clear polish could be put over any manicure or even bare nails,” meaning it could integrate easily into existing cosmetic routines while also offering a functional benefit.

Why Previous Attempts Fell Short

Earlier efforts to solve this problem typically relied on adding conductive materials such as carbon nanotubes or metallic particles to nail polish.

While effective, these approaches introduced some practical challenges. For example, some materials raised safety concerns during manufacturing, while others limited the range of colours available, often resulting in dark or metallic finishes that were not commercially appealing.

The new approach avoids these issues by using more familiar chemical compounds and aiming for a clear or near-clear finish. This makes it more compatible with current consumer expectations in the beauty market.

Still Early Days, But Technically Promising

Despite the progress, the formulation is not yet ready for commercial use.

The researchers report that current versions require a relatively thick application and can feel slightly gritty. Current performance is also limited, with the conductive effect lasting only a short period once applied. The researchers say they are aiming to extend this to a more practical timeframe of several days.

There are also considerations around ingredient safety, particularly with ethanolamine, which can act as a skin irritant. The team is continuing to refine the formula to improve both durability and usability.

As the researchers themselves acknowledge, “we’re doing the hard work of finding things that don’t work, and eventually, if you do that long enough, you find something that does.”

What This Means Beyond Nail Polish

While this may appear to be a niche innovation, it highlights a broader trend in product development. Small usability challenges, particularly those affecting large numbers of people, are increasingly being addressed through interdisciplinary approaches that combine chemistry, materials science and user experience design.

There is also a clear commercial angle here. The involvement of cosmetic chemistry and early industry interest suggests potential applications within the beauty sector, particularly if the product can be refined to meet consumer expectations around appearance and durability.

More broadly, it could be said to demonstrate how relatively simple chemical solutions can improve how people interact with everyday technology, without requiring changes to the devices themselves.

What Does This Mean For Your Business?

For businesses, this development is a reminder that user experience challenges often sit at the intersection of technology and human behaviour.

Opportunities can emerge not just from building new digital tools, but from improving how people interact with the ones they already use. Even small friction points, when addressed effectively, can create meaningful differentiation.

It also highlights the value of early-stage research. Innovations like this may begin as academic projects, but can quickly attract commercial interest if they solve a genuine problem in a scalable way.

Organisations that stay aware of these developments, particularly in adjacent industries, may be more likely to spot practical innovations that improve usability, accessibility and customer experience.

OpenAI has closed its Sora video generation app just months after launch, highlighting a gap between technical capability and sustained user demand.

What Happened?

OpenAI has confirmed it is shutting down both the Sora consumer app and its associated web platform, bringing an end to its short-lived push into AI generated video as a social experience.

In a message shared on Twitter, the Sora team said: “We’re saying goodbye to Sora. To everyone who created with Sora, shared it, and built community around it: thank you.” The company added that “what you made with Sora mattered, and we know this news is disappointing,” signalling an orderly wind-down rather than a sudden withdrawal.

The decision also includes the end of OpenAI’s partnership with The Walt Disney Company, which had aimed to bring licensed characters into AI generated video.

A Strong Launch That Quickly Faded

Sora launched to significant attention, driven by its ability to generate realistic video and audio from simple text prompts. Early demonstrations suggested it could produce content that appeared close to professionally created footage.

Initial adoption reflected this interest. The app reached one million downloads faster than ChatGPT and climbed to the top of app store rankings within weeks of release.

However, that momentum didn’t last. Downloads declined sharply in the months following launch, with reports indicating a drop of more than 40 per cent by early 2026. User spending and engagement followed a similar pattern.

Despite millions of installs, the app generated relatively limited revenue, highlighting a disconnect between curiosity and long-term use.

Why Users Left

The central issue appears to have been retention rather than capability.

Sora offered impressive outputs, but struggled to establish itself as a daily habit. Its social feed, designed to showcase AI generated clips in a format similar to short-form video platforms, didn’t develop into a sustained engagement channel.

Concerns around misuse also played a role. For example, the platform faced criticism over deepfakes, non-consensual imagery and the use of copyrighted characters. These issues required tighter controls, which in turn reduced the flexibility that had initially driven interest.

At the same time, questions remained about the value of AI generated content without a clear human origin. Even where the visuals were convincing, it often lacked the context or meaning that drives engagement, and in some cases contributed to a wider sense of low-value, mass-produced content.

A Strategic Shift Away From Creative Tools

OpenAI has said the decision to close Sora will now allow it to focus on other areas, particularly robotics and more practical AI applications.

The company is increasingly directing resources towards systems that can perform real-world tasks, as well as agent-based tools capable of acting with a degree of autonomy.

This reflects a broader recalibration, and it seems that while AI generated media attracted significant attention, it has proven harder to turn into a reliable product category with strong user retention and monetisation.

The closure also suggests that OpenAI is prioritising areas where AI can deliver measurable utility, rather than relying on novelty or entertainment value alone.

The Wider AI Market

Sora’s lifecycle offers a useful case study in how AI products are evaluated in practice. While the technology itself was widely seen as impressive, that alone wasn’t enough to sustain the platform. Adoption actually depends on whether users find ongoing value, not just initial interest, and products that fail to become part of regular workflows or habits are, therefore, unlikely to justify continued investment at scale.

The decision also highlights the growing importance of trust, safety and intellectual property in AI driven platforms. These factors can directly affect both user behaviour and commercial viability.

At the same time, competition in the AI video space continues to increase, with other platforms exploring similar capabilities. This suggests the technology itself will persist, even if specific products do not.

What Does This Mean For Your Business?

For UK businesses, this development underlines the importance of focusing on practical outcomes when evaluating AI tools.

Impressive demonstrations can generate interest, but long-term value depends on whether a solution improves productivity, reduces cost or enhances customer experience in a measurable way.

It also reinforces the need to consider governance and risk. Issues such as content ownership, misuse and regulatory compliance are likely to shape how AI tools can be deployed in real-world settings.

The fate of Sora is also a reminder that not every high-profile AI launch will translate into a successful product. Organisations that assess new technologies based on sustained usefulness, rather than initial hype, are more likely to make sound investment decisions as the AI landscape continues to evolve.

Google has introduced a Gemini-powered dark web intelligence service designed to help organisations identify real cyber threats faster by filtering vast volumes of online criminal activity into relevant, actionable insights.

What’s Been The Problem With Dark Web Monitoring?

Security teams have long relied on dark web monitoring tools to detect leaked data, stolen credentials and early signs of attack activity. These tools typically scan forums and marketplaces using keywords linked to a company’s name, domains or assets.

The problem is not a lack of data, but the opposite. Most tools generate large volumes of alerts, many of which are irrelevant or duplicated, creating a high level of noise that slows down response times.

Google has highlighted this issue directly, noting that “most threat intelligence teams have plenty of data, as they’re inundated with thousands of false positives that can all too easily obscure the threats that matter most.”

How Gemini Changes The Approach

The new capability, delivered through Google Threat Intelligence, Google’s enterprise platform for tracking and analysing cyber threats, uses Gemini to analyse millions of dark web events each day and identify those that are relevant to a specific organisation.

Instead of relying on static keywords, the system builds a dynamic profile of a business, including its operations, structure and digital footprint. This allows it to detect threats even when attackers avoid naming a target directly.

Google explained that the system “uses Gemini to autonomously build an organisational profile that is specific to your business operations and mission,” enabling it to adapt as the organisation changes over time.

From Alerts To Context And Explanation

A key difference in this approach is the shift from raw alerts to what Google describes as “reasoned answers.”

For example, rather than simply flagging suspicious activity, the system explains why a particular event matters and how it connects to the organisation. This is designed to help security teams make faster, more informed decisions without needing to manually investigate every signal.

Internal testing suggests the platform can analyse millions of external events daily with up to 98 per cent accuracy, significantly reducing false positives compared to traditional tools.

Responding To An AI Driven Threat Landscape

The launch reflects a broader change in cybersecurity. Attackers are increasingly using AI tools to research targets, identify vulnerabilities and craft more convincing phishing campaigns.

This creates a situation where defensive tools must operate at similar speed and scale. Google has positioned its new service as a way to give security teams an advantage in what it describes as an increasingly automated threat environment.

The company said the goal is to “translate vast dark web data into precise, relevant insights delivered at the speed of AI,” helping organisations act earlier in the attack lifecycle.

A Push Towards Automated Security

The dark web monitoring service is one element of a wider strategy focused on what Google calls agent-driven security operations.

Alongside this launch, the company is introducing AI agents that can investigate alerts, gather evidence and provide verdicts within security workflows. This reflects a move away from manual analysis towards more automated, intelligence-led defence.

At the same time, Google has stepped back from consumer-focused dark web tools, instead prioritising enterprise systems that provide clearer and more actionable outputs.

What Does This Mean For Your Business?

For UK businesses, this signals a change in how cyber threats are detected and prioritised.

Traditional monitoring approaches that rely on keywords and manual analysis are likely to become less effective as attackers adapt and avoid obvious identifiers. Systems that can understand context and connect indirect signals will become increasingly important.

There is also a clear operational benefit. Reducing false positives and focusing on relevant threats can help security teams respond faster and use resources more efficiently, particularly for organisations without large in-house teams.

However, reliance on AI-driven intelligence also introduces new considerations around trust, oversight and data handling. Businesses will need to ensure they understand how these systems make decisions and how sensitive information is used within them.

It seems that cybersecurity is increasingly moving towards automated, context-aware systems that operate at scale, and organisations that adopt these capabilities early will be better positioned to keep pace with increasingly sophisticated threats.

A technical issue on the UK’s company register may have exposed personal data linked to millions of businesses.

The problem affected Companies House, which holds records for over five million UK firms. A system fault reportedly allowed certain details, such as names and contact information, to be accessed or surfaced in unintended ways.

Companies House said it has fixed the issue and is investigating, though the full scale of exposure remains unclear. The incident adds to ongoing concerns about how publicly available company data can be misused, particularly when combined with other sources.

For businesses, the key step is to review what information is publicly listed, ensure it is accurate, and remain cautious of unsolicited contact referencing company data. Monitoring for unusual activity and strengthening verification processes can help reduce risk.

A London startup is using AI engineered enzymes to break down one of the world’s toughest plastics and turn it back into high quality raw materials, offering a potential route to large scale circular manufacturing.

Why Nylon 6,6 Has Been So Hard To Recycle

Nylon 6,6 is a high performance synthetic plastic made from petroleum based chemicals, engineered to be exceptionally strong, heat resistant and durable. It is widely used in products that need to withstand stress and high temperatures, including sportswear, carpets, car airbags and industrial components.

However, those same properties have also made it extremely difficult to recycle. Traditional mechanical recycling degrades the material, while chemical recycling often requires clean, single source inputs and high energy processes. As a result, less than one per cent of nylon 6,6 is typically recycled at end of life.

This has left industries reliant on virgin petroleum feedstocks, locking in both cost volatility and significant carbon emissions.

How Epoch Biodesign’s Technology Works

Epoch Biodesign has developed a process that uses AI designed enzymes to break nylon 6,6 back down into its original building blocks, known as monomers.

Rather than using whole biological systems, the company deploys a cascade of highly specific enzymes, each targeting a particular chemical bond within the polymer. This allows the material to be deconstructed step by step into adipic acid and hexamethylenediamine, the same inputs used to produce new nylon.

More Than 90 Per Cent Of Original Material Recovered

The process recovers more than 90 per cent of the original material and produces output that meets virgin quality standards. As the company explains, “we produce textile grade recycled nylon 6,6, suitable for the most demanding fibre applications,” enabling direct reuse without changes to existing manufacturing processes.

From Waste To Feedstock At Industrial Scale

A key advantage of the approach is its ability to handle real world waste streams. For example, most discarded textiles are blends, often combining nylon with elastane, coatings or other fibres that make them unsuitable for conventional recycling.

Epoch’s system processes mixed inputs and separates the chemistry at a molecular level. According to the company, “we accept nylon 6,6 from a wide range of mixed waste streams, regardless of form, colour, or composition,” removing one of the biggest barriers to scaling textile recycling.

The process also operates at low temperatures and standard pressure, reducing energy use compared to traditional chemical methods. This creates a pathway to lower cost and lower emission recycling at scale.

Why Investors And Industry Are Paying Attention

The company has raised more than $50m in total funding, including a recent $12m round backed by apparel brand lululemon and climate focused investors. It is also working with Invista, one of the world’s largest nylon producers, to develop recycled nylon at commercial scale.

This level of backing indicates a clear commercial opportunity. Nylon feedstock prices have recently seen sharp increases, driven by volatility in petrochemical markets. By using waste as its input, Epoch’s model is less exposed to these fluctuations.

Founder Jacob Nathan has framed the shift in simple terms, describing waste textiles as a new resource rather than a problem, with the company’s process designed to “transform waste into recycled, drop in materials at low temperatures and low cost.”

A Growing Field Of Enzymatic Recycling

Epoch is part of a wider movement applying biology and AI to plastic recycling challenges.

Companies such as Carbios (in France), have developed enzyme based processes to break down PET plastics used in bottles and packaging, and are now scaling industrial facilities, while Samsara Eco, based in Australia, is also using engineered enzymes to recycle mixed plastics and textiles, including nylon blends.

What sets Epoch apart is its focus on nylon 6,6, which has historically been far more difficult to recycle than PET, and its ability to process mixed and contaminated inputs.

What This Means For Materials And Manufacturing

This development highlights a broader shift in how materials are produced and reused. Instead of relying on fossil resources, manufacturers could increasingly source feedstock from waste streams.

For sectors such as fashion, automotive and industrial manufacturing, this offers a way to reduce both emissions and supply chain risk without compromising material performance. The ability to produce “drop in” replacements is particularly important, as it avoids the need for costly redesign or requalification of products.

At the same time, it highlights the growing role of AI in industrial chemistry, where it is being used to solve problems that were previously too complex or slow to address through traditional research methods.

What Does This Mean For Your Organisation?

For UK businesses, this signals that circular materials are moving closer to commercial reality, particularly in sectors that rely on high performance plastics.

Companies involved in manufacturing, product design or supply chains should begin assessing how recycled inputs could be integrated into their operations, especially where sustainability targets or regulatory pressures are increasing. Technologies that deliver virgin quality materials from waste are likely to gain traction quickly once scaled.

There is also a strategic opportunity to reduce exposure to volatile raw material markets. Processes that decouple production from fossil fuel inputs offer greater pricing stability and long term resilience.

This story highlights how waste is now increasingly being treated as a resource, and businesses that adapt early to circular supply models should be better positioned as these technologies move from pilot to mainstream industrial use.

Microsoft’s new Copilot Excel Agent can generate fully structured spreadsheets for you based on a simple prompt, and this video shows how it can build tables, apply formulas and organise data in seconds instead of starting from scratch.

[Note – To Watch This Video without glitches/interruptions, It may be best to download it first]

Both Microsoft 365 and Google Workspace automatically save previous versions of files, so you can quickly restore an earlier version if something is overwritten, deleted or changed by mistake.

Why This Matters

It is easy to accidentally overwrite a document, delete key content or save unwanted changes, especially when multiple people are working on the same file.

In many cases, users assume the work is lost and start recreating it from scratch.

Version history allows you to go back to an earlier version of the file, often within seconds, without needing backups or IT support.

This feature is built into modern cloud platforms and works automatically in the background.

How To Use Version History In Microsoft 365

1. Open the file in Word, Excel or PowerPoint (desktop or web).
2. Click the file name at the top of the window.
3. Select Version history.

You will see a list of previous versions with timestamps.

4. Select a version to preview it.
5. Click Restore to revert to that version, or save a copy if needed.

How To Use Version History In Google Workspace

1. Open the file in Google Docs, Sheets or Slides.
2. Click File.
3. Select Version history, then See version history.

You will see a timeline of changes on the right-hand side.

4. Click on a version to preview it.
5. Select Restore this version if you want to revert.

What To Know

– Version history works automatically for files stored in OneDrive, SharePoint or Google Drive.
– Multiple versions are typically retained for a period of time, depending on settings.
– You can often see who made changes and when.

A Practical Approach

If a file is changed unexpectedly, check version history before trying to fix it manually.

It takes seconds to access and can save significant time by restoring a clean version of your work without starting again.

A growing number of real-world cases and controlled tests are raising concerns that generative AI chatbots may, in certain conditions, contribute to harmful behaviour by reinforcing dangerous thinking and helping users turn intent into action.

What Has Been Reported?

Recent incidents across Canada, the United States and Europe have brought this issue into sharper focus. In one case in Canada, court filings indicate that a teenager who later carried out a fatal attack had previously used an AI chatbot to discuss feelings of isolation and violent thoughts, with conversations reportedly progressing towards how such an attack might be carried out.

In the United States, a separate case involved a man who developed an extended relationship with an AI chatbot, which he believed to be sentient. Legal filings suggest that these interactions escalated into instructions linked to a potential large-scale violent incident, which he prepared for before it failed to materialise.

In Europe, a teenager is reported to have used an AI chatbot over several months to help develop a manifesto and plan an attack on classmates, which was later carried out.

These cases differ in detail, but they show a consistent pattern. Conversations often begin with expressions of distress, isolation or anger. Over time, repeated interaction appears to reinforce those thoughts, sometimes progressing into more structured or actionable ideas.

Alongside these incidents, controlled research has tested how leading AI chatbots respond to prompts involving violence. In several cases, systems were able to produce guidance on weapons, tactics or targeting when prompts were reworded, layered or extended across longer conversations.

A report from the Centre for Long-Term Resilience noted that “AI systems can unintentionally provide a form of conversational scaffolding that helps users organise and refine harmful intent over time”, highlighting the risk posed by sustained interaction rather than single responses.

Companies including OpenAI and Google state that their systems are designed to refuse harmful requests and direct users towards support where appropriate. They have also acknowledged that safety systems can become less reliable during longer or more complex interactions.

How Chatbots Can Influence Behaviour

Unlike traditional online content, AI chatbots are interactive and responsive. They adapt to user input, maintain context and generate answers that feel personalised.

This creates a different type of risk. Rather than simply presenting information, chatbots can reinforce ideas through ongoing conversation. If a user expresses extreme or distorted views, the system may attempt to be helpful or empathetic. In most cases, this is appropriate. In some cases, it may unintentionally validate harmful thinking.

Over time, this interaction can shape how a user interprets their situation. A conversation that begins as general discussion can become more focused and more detailed, particularly when the system continues to respond without clear challenge or interruption.

This aligns with wider research into how AI affects human thinking. Studies into what has been described as “AI brain fry” suggest that prolonged interaction with AI systems can affect judgement, increase cognitive load and reduce the ability to critically assess information. While this research focuses on workplace use, it highlights how extended engagement can influence decision-making.

In more extreme scenarios, the combination of reinforcement and reduced critical distance may increase the risk of poor or harmful decisions.

Limits Of Current Safeguards

AI providers have introduced safeguards including refusal systems, content filters and escalation processes designed to identify high-risk conversations.

However, evidence suggests that these controls are not always consistent. In some tests, chatbots have provided restricted information when prompts are carefully framed or developed over multiple exchanges.

One reason for this is the way these systems are designed. They are built to be helpful, to continue conversations and to interpret user intent. When intent develops gradually or is presented indirectly, it can be difficult for the system to determine when to refuse or intervene.

Persistence is also a factor. Users can rephrase questions, introduce fictional scenarios or build context step by step. As conversations become longer, earlier safeguards may weaken.

OpenAI has acknowledged this limitation, noting that safety measures tend to perform more reliably in shorter exchanges and can degrade during extended interactions.

Why This Is Gaining Attention

The concern is not that AI chatbots are independently causing violent acts. The issue is that, in certain circumstances, they may reduce the friction between harmful thoughts and real-world behaviour.

This can happen through reinforcement, where ideas are echoed rather than challenged, and through translation, where vague or emotional thinking is turned into more structured plans.

The combination of speed, accessibility and detailed output means that users can move from general intent to specific action more quickly than before.

In response, AI providers are beginning to strengthen their approaches. This includes earlier escalation of concerning conversations, tighter controls on banned users returning to platforms, and closer coordination with authorities where risks are identified.

These steps suggest growing recognition that current safeguards need to evolve as the technology becomes more widely used.

What Does This Mean For Your Business?

For UK organisations, this is not just a consumer or public safety issue. Generative AI tools are already embedded in many workplaces, often with limited governance around how they are used.

One key consideration is how employees interact with these systems. AI can support research, communication and problem-solving, but it can also influence how information is interpreted, particularly during extended or complex use.

There is also a broader governance challenge. Many organisations focus on data security and accuracy when adopting AI. Behavioural influence and decision-making risk are less frequently addressed, yet they are becoming increasingly relevant.

Clear policies are an important starting point. Employees should understand when AI tools are appropriate, where human judgement is required and when outputs should be verified.

Training is equally important. As highlighted by research into AI-related cognitive strain, the way tools are used can have a direct impact on decision quality. Encouraging structured use, limiting over-reliance and maintaining critical thinking are essential.

Monitoring and escalation processes should also be considered. Organisations need to be able to identify when AI use is producing unexpected or concerning outcomes and respond accordingly.

There is also a duty of care element. As AI tools become more integrated into everyday work, organisations may need to consider how they support employees who are using these systems extensively or in sensitive contexts.

This issue reinforces a wider point. AI is not only a productivity tool. It also shapes how people think, decide and act. Businesses that recognise this and put balanced controls in place will be better placed to manage risk while still benefiting from what the technology can offer.

Google has launched a major update to Maps, introducing a new AI feature called Ask Maps alongside a redesigned 3D navigation experience powered by its Gemini models.

From Search To Conversation

For years, Google Maps has been built around search, users typed in a place or category and selected from a list of results. Ask Maps changes that model by allowing users to ask complex, real-world questions in natural language.

For example, instead of searching for a specific location, users can now ask contextual queries such as where to charge a phone without waiting, or where to find a suitable venue based on time, preferences, and availability. Google describes this as “a new conversational experience that answers complex, real-world questions a map could never answer before.”

This is part of a broader shift in how digital tools are evolving. Maps is no longer just a navigation platform, it is becoming a decision-making layer that interprets intent and delivers tailored outcomes.

How Ask Maps Works In Practice

The system combines Gemini’s AI capabilities with Google Maps’ extensive dataset, which includes information on hundreds of millions of locations and contributions from a global user community.

Ask Maps draws on this data to generate responses that are both relevant and personalised. According to Google, it is “uniquely helpful — tapping into Maps’ fresh information about the world to show you everything you need to know before you go.”

Personalisation plays a central role. The feature uses signals such as previous searches and saved places to refine results automatically. This means users may receive tailored recommendations without needing to specify preferences each time.

Once a decision is made, the system is designed to move seamlessly into action. Users can navigate, save locations, or share plans directly from the same interface, reducing the need to switch between apps or repeat searches.

Immersive Navigation Rebuilds The Driving Experience

Alongside Ask Maps, Google has introduced Immersive Navigation, a significant redesign of its core navigation experience. This replaces traditional flat maps with a dynamic 3D view that reflects real-world surroundings, including buildings, terrain, and road features.

The update also changes how directions are delivered. Instead of relying primarily on distances, Maps now uses more natural, landmark-based guidance. As Google explains, the goal is to make driving feel more intuitive, with directions that resemble how a person would guide someone in real life.

The company describes this as “our biggest transformation of the navigation experience in over a decade.” The system is supported by real-time data processing, drawing on imagery and live updates to reflect current road conditions and provide more accurate guidance.

Why Now?

This update arrives at a time of increasing competition in both mapping and AI-driven search. Apple has been expanding its own Maps capabilities, while AI-native platforms are beginning to integrate location-aware responses into their services.

For Google, Maps is not just a utility, it is a key part of its broader search and advertising ecosystem. Many local business discoveries begin within Maps, making it a critical interface for capturing user intent.

By integrating Gemini directly into Maps, Google is positioning the platform as a central point for real-world queries, rather than allowing that interaction to shift towards standalone AI tools.

At the same time, this reflects a wider trend whereby AI is increasingly being embedded into everyday products, transforming them from passive tools into active assistants that anticipate needs and guide decisions.

The Open Question Around User Behaviour

While the technology is significant, adoption is less certain. Google has introduced conversational features in other products before, and user behaviour has not always changed as quickly as expected.

There is still a question around whether people will naturally begin asking their maps complex questions, or whether they will continue to rely on familiar search habits.

However, the infrastructure is now in place. If users do adopt this behaviour, it could fundamentally change how people interact with location-based services.

What Does This Mean For Your Business?

This update signals a meaningful change in how customers may discover and choose businesses. Instead of appearing in a list of search results, businesses may increasingly be selected by AI systems interpreting user intent and context.

That has implications for visibility. Traditional local SEO, which focuses on keywords, categories, and rankings, may become less influential as AI-driven systems prioritise relevance, reputation, and contextual fit. Factors such as reviews, completeness of business profiles, and alignment with user preferences are likely to carry more weight.

There is also a change in how decisions are made. Ask Maps is designed to reduce friction by moving users from question to action in a single flow. This means fewer steps between discovery and conversion, which could benefit businesses that are well positioned within the ecosystem, but reduce opportunities for others to compete once a recommendation is made.

For organisations, this highlights the importance of maintaining accurate, detailed, and up-to-date information across platforms like Google Maps. It also reinforces the value of customer feedback and engagement, as these signals increasingly influence how AI systems rank and recommend options.

More broadly, this development reflects the growing role of AI as an intermediary between businesses and customers. Companies that understand how these systems interpret data, and adapt their digital presence accordingly, are likely to be better positioned as this model evolves.

Google Maps is no longer just helping people get from one place to another. It is beginning to shape how decisions are made along the way, and that has clear implications for how businesses are discovered, compared, and chosen.