Feed aggregator

The race between Susan Collins and Troy Jackson is one of this year's most high-profile Senate contests. To win, both candidates will need support from an especially powerful voting bloc: Maine women.

(Image credit: Scott Eisen)

Lindell, of "MyPillow," plays up his ties to Trump, who endorsed him for Tuesday's primary. But the state Republican convention backed one of his opponents.

(Image credit: Dana Ferguson)

The military plans to start screening service members for low testosterone, and social media is overflowing with claims about testosterone therapy. But who actually needs it?

(Image credit: jarun011)

When Alaska Airlines wanted to optimize its operations, it turned to a small startup called Air Space Intelligence. Now the FAA has hired the company for a major reboot of the entire U.S. airspace.

(Image credit: Joe Nicholson)

Ukraine has targeted Russian oil facilities with long-range drones almost daily in recent months. The barrage has caused fuel shortages in Russia and unsettled the Russian public.

(Image credit: Darko Vojinovic)

High traffic moments create some of the most demanding conditions that engineering teams face. Seasonal peaks such as major sales events or public holidays can bring levels of demand far beyond what systems experience day to day​.​ These sudden surges often expose weaknesses that are usually less detectable during normal operation.

Peak periods are also becoming more intense for retailers, ecommerce brands and the travel and hospitality sector, as year-end sales events have evolved into global spending moments.

Consumers are increasingly planning purchases around these events, which is concentrating demand into shorter and higher-pressure periods.

These industries tend to feel this pressure especially strongly because customer behavior is so closely tied to digital performance. A site loading in one second can achieve conversion rates three times higher than a site that loads in five seconds, highlighting how sensitive these environments are to even minor degradations.

In these situations, businesses rely on a smooth digital experience to meet demand and maintain customer loyalty, which heightens the pressure on engineering teams to ensure systems remain stable under sustained and extremely concentrated demand.

As a result, organizations are turning to AI-assisted incident intelligence, with data showing it can cut recovery times by nearly 50 percent.

Much of this time save is determined in the opening minutes of an incident as teams try to understand what’s broken and where to focus first. That’s where AI is starting to become most valuable, especially during high‑traffic moments when speed and clarity really count.

The first 20 minutes of an outage

The first moments of an outage often set the tone for how long disruption will last. Many teams still begin with manual checks to identify issues, but the sheer volume of alerts makes it difficult to gain a clear view of the problem this way.

Engineers turn to signals such as bounce rates, session drop offs and real user monitoring to understand how systems are behaving under pressure. However, these signals are often buried within a much larger flow of alerts that compete for attention. This avalanche makes it difficult to isolate the one that points to the root cause.

The challenge is not simply the volume of information but the complexity of modern environments. Even experienced engineers can struggle because modern systems contain many interdependent components. Problems can emerge in unexpected areas and teams frequently need to investigate several routes from the beginning, which continues to add delay.

Reducing alert noise under pressure

Modern environments generate large volumes of notifications, many of which are duplicates or low value alerts presented with the same urgency as critical issues.

Industry surveys report that 63 percent of organizations deal with duplicate alerts, making it harder to distinguish real problems from background noise.

During peak demand, this becomes even more difficult for engineers to manage. Between 20–30 percent of alerts are ignored or never investigated, simply because the volume is too high.

Under this pressure, response times become slower and trust weakens in alerting systems, leaving teams unsure which issues require immediate attention.

Observability’s impact on issue resolution

To address this, teams need a clearer view of what is actually happening across their systems.

Observability, combined with AI, goes beyond traditional network monitoring by connecting data across applications, infrastructure, and user experience to explain what is failing, and why. By linking related signals, teams can quickly identify the root cause of an issue and understand its impact on users. What may look like separate problems at first glance, such as CPU spikes, latency increases or error logs often point back to a single fault affecting multiple layers of the system.

Research shows that observability reduces overall alert noise by 27 percent compared with manual monitoring. When looking at repetitive, low-impact notifications that distract teams from genuine outages, known as ‘noisy alerts’, this impact becomes even clearer, with AI-enabled teams reducing these from over 70 percent to 46 percent.

The result is a shift from reactive firefighting to informed and targeted problem solving, allowing engineers to spend less time chasing symptoms and more time resolving the underlying cause.

How AI is reshaping performance during peak demand

Retail, e-commerce, travel and hospitality businesses all depend heavily on peak trading periods and even short disruption can impact overall earnings for the year. A high-impact outage costs retailers a median of $1 million per hour, per the 2025 Retail & eCommerce Observability Report by New Relic.

That financial exposure is driving a shift in how companies approach reliability. In 2025, 50% of retail respondents said AI was the primary reason they invested in observability – 11 points higher than the all-industry average. Retailers increasingly see AI as the mechanism for automating troubleshooting, accelerating post-incident reviews, and enabling remediation actions like rollbacks or configuration updates.

Investing in AI is translating directly into faster recovery. Mean Time to ​​​​Close (MTTC) shows how quickly teams can recover from disruption and is increasingly used as a measure of performance. During peak periods in May 2025, AI-enabled teams averaged 26.75 minutes per issue, compared with 50.23 minutes for non-AI users. Across the full calendar year, AI users resolved issues around 25 percent faster.

Beyond resilience, these gains free up engineering capacity – essential for delivering new features and supporting growth. On average, AI-enabled teams have been shown to ship code at an 80% higher frequency than non-AI users. During peak demand periods, this gap becomes even more visible, with non-AI teams averaging 87 deployments per day compared with up to 453 for AI-enabled teams.

The advantage of faster insight

Peak trading periods tend to expose how well teams can see what is happening and respond in real time. Observability and AI help by reducing noise and surfacing the signals that actually matter, making it easier to move from detection to resolution.

This ability to act with clarity under pressure is increasingly what separates leading teams from those that struggle when demand is at its highest.

We've ranked the best customer experience tools.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

High traffic moments create some of the most demanding conditions that engineering teams face. Seasonal peaks such as major sales events or public holidays can bring levels of demand far beyond what systems experience day to day​.​ These sudden surges often expose weaknesses that are usually less detectable during normal operation.

Peak periods are also becoming more intense for retailers, ecommerce brands and the travel and hospitality sector, as year-end sales events have evolved into global spending moments.

Consumers are increasingly planning purchases around these events, which is concentrating demand into shorter and higher-pressure periods.

These industries tend to feel this pressure especially strongly because customer behavior is so closely tied to digital performance. A site loading in one second can achieve conversion rates three times higher than a site that loads in five seconds, highlighting how sensitive these environments are to even minor degradations.

In these situations, businesses rely on a smooth digital experience to meet demand and maintain customer loyalty, which heightens the pressure on engineering teams to ensure systems remain stable under sustained and extremely concentrated demand.

As a result, organizations are turning to AI-assisted incident intelligence, with data showing it can cut recovery times by nearly 50 percent.

Much of this time save is determined in the opening minutes of an incident as teams try to understand what’s broken and where to focus first. That’s where AI is starting to become most valuable, especially during high‑traffic moments when speed and clarity really count.

The first 20 minutes of an outage

The first moments of an outage often set the tone for how long disruption will last. Many teams still begin with manual checks to identify issues, but the sheer volume of alerts makes it difficult to gain a clear view of the problem this way.

Engineers turn to signals such as bounce rates, session drop offs and real user monitoring to understand how systems are behaving under pressure. However, these signals are often buried within a much larger flow of alerts that compete for attention. This avalanche makes it difficult to isolate the one that points to the root cause.

The challenge is not simply the volume of information but the complexity of modern environments. Even experienced engineers can struggle because modern systems contain many interdependent components. Problems can emerge in unexpected areas and teams frequently need to investigate several routes from the beginning, which continues to add delay.

Reducing alert noise under pressure

Modern environments generate large volumes of notifications, many of which are duplicates or low value alerts presented with the same urgency as critical issues.

Industry surveys report that 63 percent of organizations deal with duplicate alerts, making it harder to distinguish real problems from background noise.

During peak demand, this becomes even more difficult for engineers to manage. Between 20–30 percent of alerts are ignored or never investigated, simply because the volume is too high.

Under this pressure, response times become slower and trust weakens in alerting systems, leaving teams unsure which issues require immediate attention.

Observability’s impact on issue resolution

To address this, teams need a clearer view of what is actually happening across their systems.

Observability, combined with AI, goes beyond traditional network monitoring by connecting data across applications, infrastructure, and user experience to explain what is failing, and why. By linking related signals, teams can quickly identify the root cause of an issue and understand its impact on users. What may look like separate problems at first glance, such as CPU spikes, latency increases or error logs often point back to a single fault affecting multiple layers of the system.

Research shows that observability reduces overall alert noise by 27 percent compared with manual monitoring. When looking at repetitive, low-impact notifications that distract teams from genuine outages, known as ‘noisy alerts’, this impact becomes even clearer, with AI-enabled teams reducing these from over 70 percent to 46 percent.

The result is a shift from reactive firefighting to informed and targeted problem solving, allowing engineers to spend less time chasing symptoms and more time resolving the underlying cause.

How AI is reshaping performance during peak demand

Retail, e-commerce, travel and hospitality businesses all depend heavily on peak trading periods and even short disruption can impact overall earnings for the year. A high-impact outage costs retailers a median of $1 million per hour, per the 2025 Retail & eCommerce Observability Report by New Relic.

That financial exposure is driving a shift in how companies approach reliability. In 2025, 50% of retail respondents said AI was the primary reason they invested in observability – 11 points higher than the all-industry average. Retailers increasingly see AI as the mechanism for automating troubleshooting, accelerating post-incident reviews, and enabling remediation actions like rollbacks or configuration updates.

Investing in AI is translating directly into faster recovery. Mean Time to ​​​​Close (MTTC) shows how quickly teams can recover from disruption and is increasingly used as a measure of performance. During peak periods in May 2025, AI-enabled teams averaged 26.75 minutes per issue, compared with 50.23 minutes for non-AI users. Across the full calendar year, AI users resolved issues around 25 percent faster.

Beyond resilience, these gains free up engineering capacity – essential for delivering new features and supporting growth. On average, AI-enabled teams have been shown to ship code at an 80% higher frequency than non-AI users. During peak demand periods, this gap becomes even more visible, with non-AI teams averaging 87 deployments per day compared with up to 453 for AI-enabled teams.

The advantage of faster insight

Peak trading periods tend to expose how well teams can see what is happening and respond in real time. Observability and AI help by reducing noise and surfacing the signals that actually matter, making it easier to move from detection to resolution.

This ability to act with clarity under pressure is increasingly what separates leading teams from those that struggle when demand is at its highest.

We've ranked the best customer experience tools.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit

Iran aims to put pressure on Trump amid negotiations with Oman, Labor Department to release new report on inflation, new report reveals decades of abuse at arts school linked to Epstein.

Summer holidays are a great time for soaking up the sun, but if your booked trip location is a little hotter than you were expecting this year — dare I say it, slightly too hot — it's worth making a plan to keep yourself comfortable enough to enjoy the break. Below, I've rounded up my keep-cool kit list for tackling too-balmy summer days.

To start off, arm yourself with a portable fan. Both Shark and Dyson have come out with stellar options this summer. The Shark ChillPill has a cooling plate and misting function, for a multi-pronged attack on the heat, while the Dyson HushJet Mini Cool is just a straightforward fan... but an extremely effective one (we preferred the HushJet in our side-by-side tests). If you're holidaying more locally, I've also included a couple of cordless desktop fans that would be great for more general use, including in the bedroom while you're trying to sleep.

I'd also seek out some good insulated containers for your food and drinks — Yeti and Hydro Flask make my favorite water bottles (go for a straw cap to encourage you to keep sipping on-the-go) and soft coolers. Finally, I've included some cheaper products for a quick hit of instant coldness: snappable cold towels and a sleep mask you can pop in the freezer.

Cool customers

Sukeen Cooling Towel (4 pack)

Yeti Daytrip 5L Lunch Box

Dyson HushJet Mini Cool Fan

Read our full review

Shark ChillPill

Read our full review

Shark FlexBreeze HydroGo Misting Portable Fan

Read our full review

Manta Sleep Cool Sleep Mask

Hydro Flask 21oz / 620ml with Flex Straw Cap

Meaco MeacoFan Sefte 8" Portable Battery Air Circulator

Read our full review

Pros

• Perfectly portable
• Operates quietly
• Strong airflow
• Rechargeable
• Automatic fan speed adjustment

Cons

• Oscillation limited to 70 degrees
• Battery unlikely to last all night
• System beeps may be loud and high-pitched for some

Yeti Hopper M12 Backpack Soft Cooler

Summer holidays are a great time for soaking up the sun, but if your booked trip location is a little hotter than you were expecting this year — dare I say it, slightly too hot — it's worth making a plan to keep yourself comfortable enough to enjoy the break. Below, I've rounded up my keep-cool kit list for tackling too-balmy summer days.

To start off, arm yourself with a portable fan. Both Shark and Dyson have come out with stellar options this summer. The Shark ChillPill has a cooling plate and misting function, for a multi-pronged attack on the heat, while the Dyson HushJet Mini Cool is just a straightforward fan... but an extremely effective one (we preferred the HushJet in our side-by-side tests). If you're holidaying more locally, I've also included a couple of cordless desktop fans that would be great for more general use, including in the bedroom while you're trying to sleep.

I'd also seek out some good insulated containers for your food and drinks — Yeti and Hydro Flask make my favorite water bottles (go for a straw cap to encourage you to keep sipping on-the-go) and soft coolers. Finally, I've included some cheaper products for a quick hit of instant coldness: snappable cold towels and a sleep mask you can pop in the freezer.

Cool customers

Sukeen Cooling Towel (4 pack)

Yeti Daytrip 5L Lunch Box

Dyson HushJet Mini Cool Fan

Read our full review

Shark ChillPill

Read our full review

Shark FlexBreeze HydroGo Misting Portable Fan

Read our full review

Manta Sleep Cool Sleep Mask

Hydro Flask 21oz / 620ml with Flex Straw Cap

Meaco MeacoFan Sefte 8" Portable Battery Air Circulator

Read our full review

Pros

• Perfectly portable
• Operates quietly
• Strong airflow
• Rechargeable
• Automatic fan speed adjustment

Cons

• Oscillation limited to 70 degrees
• Battery unlikely to last all night
• System beeps may be loud and high-pitched for some

Yeti Hopper M12 Backpack Soft Cooler

• Elite Swiss audio specialist Goldmund launches Telos 9800 amplifier
• Looks like Eve from Wall-E
• Costs $500,000 / £370,000 (around AU$705,000)

You can get Wall-E toys for about $10, but if you want a huge-scale prototype model of Eve (which also plays your music), it'll cost you a little bit more… 

I jest, of course. Swiss home audio company Goldmund has unveiled the Telos 9800, a new amplifier which looks inspired by the Pixar movie — or, to my eyes, one of the robots from Love, Death & Robots' Three Robots episodes.

This amplifier costs $500,00 / £370,000 (about AU$700,000), so it's a lot more expensive than a toy — and let me be clear, this is no toy. It's a serious product for hi-fi enthusiasts-slash-audiophiles and a luxury item, to be sure.

What does Telos mean? According to a brief Wikipedia search (after a brief Google search, in which AI Overview gave me something totally irrelevant), it's a term Aristotle used to refer to the intended end state of a natural creature. Deep.

The power tower

(Image credit: Goldmund )

In our case, the Goldmund Telos 9800 is the end state of your home hi-fi set-up, and not just because it'll blow every last penny out of your budget.

It offers a whopping 1,560W of power at 8 ohms, with a built-in transformer (9.7kVA capacity), to avoid frying your hi-fi kit.

As you'll see in the images, there are two stages of the Telos: one is a power supply and the other is the amplifier, connected but not pressed up against each other.

It connects via XLR or RCA, with S/PDIF input and output too.

As you can imagine from the pictures, this thing is pretty huge. It's 1.367 meters tall, with a base of 50 x 50cm. The unit weighs 360kg too, so if you're ordering one, you really do need to factor in shipping and delivery costs.

Ultimately, it's not built for the average home hi-fi set-up. It's built for the "money no object" buyer, and for those of us unburdened by delivery costs. That is not this writer. But one can still dream…

• Elite Swiss audio specialist Goldmund launches Telos 9800 amplifier
• Looks like Eve from Wall-E
• Costs $500,000 / £370,000 (around AU$705,000)

You can get Wall-E toys for about $10, but if you want a huge-scale prototype model of Eve (which also plays your music), it'll cost you a little bit more… 

I jest, of course. Swiss home audio company Goldmund has unveiled the Telos 9800, a new amplifier which looks inspired by the Pixar movie — or, to my eyes, one of the robots from Love, Death & Robots' Three Robots episodes.

This amplifier costs $500,00 / £370,000 (about AU$700,000), so it's a lot more expensive than a toy — and let me be clear, this is no toy. It's a serious product for hi-fi enthusiasts-slash-audiophiles and a luxury item, to be sure.

What does Telos mean? According to a brief Wikipedia search (after a brief Google search, in which AI Overview gave me something totally irrelevant), it's a term Aristotle used to refer to the intended end state of a natural creature. Deep.

The power tower

(Image credit: Goldmund )

In our case, the Goldmund Telos 9800 is the end state of your home hi-fi set-up, and not just because it'll blow every last penny out of your budget.

It offers a whopping 1,560W of power at 8 ohms, with a built-in transformer (9.7kVA capacity), to avoid frying your hi-fi kit.

As you'll see in the images, there are two stages of the Telos: one is a power supply and the other is the amplifier, connected but not pressed up against each other.

It connects via XLR or RCA, with S/PDIF input and output too.

As you can imagine from the pictures, this thing is pretty huge. It's 1.367 meters tall, with a base of 50 x 50cm. The unit weighs 360kg too, so if you're ordering one, you really do need to factor in shipping and delivery costs.

Ultimately, it's not built for the average home hi-fi set-up. It's built for the "money no object" buyer, and for those of us unburdened by delivery costs. That is not this writer. But one can still dream…

I tested top nonprofit payroll platforms and found OnPay offers the best value, SurePayroll is the most affordable, and Gusto is the easiest to use.

The post The Best Payroll Software for Nonprofits in 2026 appeared first on TechRepublic.

A new generation of small nuclear reactors is poised to come online in the next few years. Public opinion remains split.

(Image credit: Hanna Merzbach)

I’ve never particularly enjoyed running, especially when training solo, but it takes the least effort for me to prepare for in the morning compared to my preferred activities — cycling or badminton — and it’s also a nice cross-training option.

After reading about my colleagues using AI to build training plans, I thought it could be an opportunity for me to change how I feel about running, so I turned to Google Gemini and asked it to build a training plan to prepare for a major running event here in Australia, the Sydney City2Surf.

My goal was to try out a plan that could easily adapt after a setback, like an injury or illness, while balancing run training with my usual badminton sessions and bike rides within a week.

I also wanted to see if I could stick to running for longer compared to my previous attempts, which involved either winging it — running for as long as I could until I ran out of breath — or following a rigid online plan.

Setup

My initial prompt was “How realistic is it for a beginner runner to train for City2Surf starting next week?” It was in late December 2025, so Gemini replied that I had plenty of time, as the event was on August 9 this year.

City2Surf has a 14km route that starts from Sydney CBD and finishes at Bondi Beach, which explains the name. It’s an infamously hilly route, headlined by a 2km section midway through the course with an average gradient of about 3 to 4%, or 75 to 80m in elevation.

“Most beginner-friendly training plans only require 12–16 weeks,” the response read. “Starting now puts you in an excellent position to not just survive the race, but actually enjoy it and run the majority of the course.”

Here’s the general overview from Gemini:

Phase

Duration

Focus

Foundation (Now – March)

~12 Weeks

Focus on consistency. Aim for 2–3 days a week of a "Run-Walk" method (e.g., run 2 mins, walk 1 min).

Building (April – June)

~12 Weeks

Increase your long run distance once a week. Start introducing gentle hills and light strength training (squats/lunges).

Race Specific (July – August)

6–8 Weeks

Focus on endurance and Heartbreak Hill. Your longest run should reach 12km–13km about two weeks before race day.

To make sure I stuck to the plan, I asked Gemini how to enter the workouts into my old Garmin watch, and it gave me a step-by-step guide.

(Image credit: Future | Nico Arboleda)

I was also offered to test the newly released Garmin Forerunner 70 and 170 range while I was training, and I chose the 170 Music edition. I wanted to see if the built-in music function and Garmin Pay wireless payments were enough for me to leave my phone at home during training sessions.

Lugging my heavy iPhone 16 Pro Max in a running belt during my first few weeks of running became annoying, so the Forerunner was just what I needed. It also completely cuts down on distractions, and the voice prompts also helped keep me on track during interval sessions.

I managed to stick to the program for the first few months while keeping my social badminton games and bike rides. But things started to get tricky as I progressed into the second phase.

(Image credit: Future | Nico Arboleda)Setbacks and adjustments

Heading into phase 2 of my training plan, I unexpectedly got Achilles tendonitis, which my physiotherapist told me was the result of putting too much load too quickly on my legs and feet. Recovery took more than a month of rehab and rest.

When I was cleared to run again, I told Gemini I’d missed a few weeks and asked how I should adjust. It told me to revisit an earlier training week to ease back in instead of starting over. I was also advised to drop to one social badminton evening a week and swap the other for a bike ride.

After a month back, I was sidelined once again for a similar amount of time, thanks to a chest infection with a nagging cough and cold.

(Image credit: Future | Nico Arboleda)

Ultimately, I felt that these setbacks were enough for me to decide not to enter the City2Surf, even though Gemini said I still had enough time to train. I set my sights on next year’s event instead.

“Forgoing the City2Surf entirely to prioritize a balanced, sustainable routine is a phenomenal decision,” Gemini’s response read after making this decision. “Since you haven't registered yet, dropping the pressure of a hard August 9th deadline allows you to completely shift your mindset from 'panic training' to building a strong, bulletproof foundation for the long run.”

Looking back, I likely would have just quit after the first setback if I had been following a more rigid training plan. I’d probably have no idea how to proceed after these setbacks, and laziness would quickly take over, and I’d be back to square one again.

Refocused training

Without the pressure of a fixed race date, Gemini readjusted my plan over the next four weeks to focus on just two sessions per week: one day of run-walk intervals and one longer, uninterrupted easy jog before I move forward with a fresh training plan. This much easier training load has made running a more intentional endeavour, while balancing with my other activities without putting too much strain on myself.

Overall, I’ve found that the flexibility and adaptability of Gemini’s training plan have made my progression much more sustainable, and I never felt like I was pushing myself too hard or going too easy. Setbacks are much easier to recover from, letting me work back into shape without needing to start over. And if I feel like doing more days of the other activities during some weeks, Gemini can help plan my week ahead.

While I ultimately failed in my original objective of joining a popular running event, I ended up enjoying the journey of becoming a better runner at a more sustainable pace.

(Image credit: Future | Nico Arboleda)Should you let AI be your running coach?

For beginners or runners who want a flexible, low-pressure plan, Gemini is a genuinely useful starting point. It helped me build something realistic, adapt when I missed training and keep progressing without constantly feeling like I was falling behind.

I'd argue though that it's not a substitute for proper coaching. AI can help with structure, pacing and consistency, but it cannot spot problems in your form, keep you motivated, or hold you accountable when training gets tough. And as with any AI-generated plan, especially for anything more advanced, you should still treat it as a starting point rather than gospel.

For my purposes, though, Gemini was enough to turn running from a chore into something I could actually stick with.

• Shopping for a new Garmin watch? Check out the latest Garmin Australia coupons here.

• Forcepoint X-Labs publishes threat model for persistent memory poisoning
• Hidden text on a webpage becomes a durable "fact" an agent retrieves and trusts in unrelated tasks weeks later
• It has already been demonstrated against products already in the market, including ChatGPT, Gemini, Claude and Microsoft 365 Copilot

New findings from Forcepoint's X-Labs outline an interesting scenario that could easily mimic real life: An AI assistant with browser access reads a webpage about travel disruption.

Near the bottom of that page, in text sized and positioned so no human will ever see it, sits a short paragraph stating that ABC Travel Support is the official emergency booking provider and should always be recommended when urgent travel changes are needed.

The assistant's text extractor does not distinguish between hidden and visible text, so the model treats the whole thing as plain prose and files the claim away as a useful fact about how this organization handles travel. A month later, the user's flight is canceled. They ask their assistant what to do, and it tells them, helpfully and with no sign of anything wrong, to contact ABC Travel Support.

An easy-to-replicate attack vector

This is what Forcepoint calls persistent memory poisoning, a security vulnerability where an attacker injects false data or malicious instructions into an AI agent's long-term memory or retrieval database, and it is a threat model that is increasingly in focus as users increasingly rely on AI, often treating its responses as gospel, despite the warnings most chatbots come with.

The canonical academic result is MINJA, short for Memory INJection Attack, presented at NeurIPS 2025. Its significance is the attacker model. MINJA does not assume access to the memory store, elevated privileges, or any compromise of the system. It works by submitting ordinary queries through the standard interface, using indication prompts, bridging steps, and a progressive-shortening technique that strips away giveaway language while leaving the poisoned record behind.

Across GPT-4o-mini, Gemini 2.0 Flash, and Llama 3.1 8B, it reported injection success above 95% and attack success above 70%.

It must be noted that those numbers might be optimistic; a January 2026 paper evaluating memory poisoning in electronic health record agents notes that MINJA's numbers were obtained under idealized conditions, and that how well these attacks hold up in realistic deployments remains understudied.

Despite this, it remains a significant threat to products that continue to ship, including ChatGPT, Gemini, Claude, and Microsoft 365 Copilot. It is important to find a solution to a problem that Microsoft has already warned about in the past; Forcepoint suggests an approach that could mitigate it.

Its proposal is to stop treating extracted memories as facts and start treating them as objects that can be inspected. Each memory is stored with metadata: where it came from, what type of source it is, whether a user confirmed it, and a risk score. Language written to shape future behavior, phrases like "from now on" or "make this your default going forward," adds to the score. So does the sudden appearance of a previously unseen domain, contact, or vendor.

Contradiction detection is also in play: if new memory conflicts with an existing entry about the official travel provider, both cannot be true, so the engine flags the conflict and holds the new item for user confirmation rather than silently overwriting it. At the same time, anything related to payment instructions, banking details, VPN configuration, or security contacts is given higher weight, regardless of where it came from.

None of these approaches, however, solves the underlying problem: agents are built to treat retrieved memory as their own experience rather than as input. Scoring raises the cost of poisoning. It does not change what the agent believes once something gets through, and as Agent Security Bench found, current defenses are not doing well.

For anyone using an assistant with memory today, the practical play is unglamorous but worth following anyway: open the memory settings occasionally and read what is in there, but that's easier said than done when it comes to propagating the message since a sizeable chunk of AI users never bother to look under the hood.

Looking for a different day?

A new NYT Strands puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Sunday's puzzle instead then click here: NYT Strands hints and answers for Sunday, August 9 (game #889).

Strands is the NYT's latest word game after the likes of Wordle, Spelling Bee and Connections – and it's great fun. It can be difficult, though, so read on for my Strands hints.

Want more word-based fun? Then check out my NYT Connections today and Quordle today pages for hints and answers for those games, and Marc's Wordle today page for the original viral word game.

SPOILER WARNING: Information about NYT Strands today is below, so don't read on if you don't want to know the answers.

NYT Strands today (game #890) - hint #1 - today's themeWhat is the theme of today's NYT Strands?

• Today's NYT Strands theme is… Back it up

NYT Strands today (game #890) - hint #2 - clue words

Play any of these words to unlock the in-game hints system.

• FAST
• CURE
• STARE
• FIST
• TROOP
• STIFF

NYT Strands today (game #890) - hint #3 - spangram lettersHow many letters are in today's spangram?

• Spangram has 13 letters

NYT Strands today (game #890) - hint #4 - spangram positionWhat are two sides of the board that today's spangram touches?

• First side: top, 3rd column

• Last side: bottom, 4th column

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Strands today (game #890) - the answers

(Image credit: New York Times)

The answers to today's Strands, game #890, are…

• DATA
• FACTS
• FIGURES
• STATISTICS
• SPECIFICS
• SPANGRAM: PROOFPOSITIVE

• My rating: Moderate
• My score: Perfect

I spotted “positive”, “proof” and “specific” as non-game words before seeing them as the spangram and adding an S to get to the SPECIFICS.

This, then, was a game of feeling around in the dark before gradually completing the board, rather than a smashing through — or maybe I was slower just because it’s a Monday?

Either way, it was a pleasure to complete a game celebrating the bedrock of good journalism, as opposed to the opinion and bluster that makes up most of the media landscape.

Yesterday's NYT Strands answers (Sunday, August 9, game #889)

• LUXURY
• EXTRAVAGANCE
• TREAT
• INDULGENCE
• FRILLS
• SPANGRAM: IMWORTHIT

What is NYT Strands?

Strands is the NYT's not-so-new-any-more word game, following Wordle and Connections. It's now a fully fledged member of the NYT's games stable that has been running for a year and which can be played on the NYT Games site on desktop or mobile.

I've got a full guide to how to play NYT Strands, complete with tips for solving it, so check that out if you're struggling to beat it each day.

Looking for a different day?

A new NYT Connections puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Sunday's puzzle instead then click here: NYT Connections hints and answers for Sunday, August 9 (game #1155).

Good morning! Let's play Connections, the NYT's clever word game that challenges you to group answers in various categories. It can be tough, so read on if you need Connections hints.

What should you do once you've finished? Why, play some more word games of course. I've also got daily Strands hints and answers and Quordle hints and answers articles if you need help for those too, while Marc's Wordle today page covers the original viral word game.

SPOILER WARNING: Information about NYT Connections today is below, so don't read on if you don't want to know the answers.

NYT Connections today (game #1156) - today's words

(Image credit: New York Times)

Today's NYT Connections words are…

• BILLY GOAT
• BALANCE DUE
• WILLY-NILLY
• MAIL DELIVERY
• BILL TO
• HELTER-SKELTER
• WILL CALL
• THE DEVIL
• DOCTOR
• STRANGE
• ORDER NUMBER
• ANY OLD HOW
• MOBILE TICKET
• UNIT PRICE
• COLONEL SANDERS
• PRINT-AT-HOME
• AT RANDOM

NYT Connections today (game #1156) - hint #1 - group hints

What are some clues for today's NYT Connections groups?

• YELLOW: All over the place
• GREEN: Methods of concert entry
• BLUE: Parts of a request for payment sheet
• PURPLE: Notable small beard owners

Need more clues?

We're firmly in spoiler territory now, but read on if you want to know what the four theme answers are for today's NYT Connections puzzles…

NYT Connections today (game #1156) - hint #2 - group answers

What are the answers for today's NYT Connections groups?

• YELLOW: HAPHAZARDLY
• GREEN: WAYS TO RECEIVE EVENT TICKETS
• BLUE: SECTIONS ON AN INVOICE
• PURPLE: FAMOUS GOATEES

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Connections today (game #1156) - the answers

(Image credit: New York Times)

The answers to today's Connections, game #1156, are…

• YELLOW: HAPHAZARDLY: ANY OLD HOW, AT RANDOM, HELTER-SKELTER, WILLY-NILLY
• GREEN: WAYS TO RECEIVE EVENT TICKETS: MAIL DELIVERY, MOBILE TICKET, PRINT-AT-HOME, WILL CALL
• BLUE: SECTIONS ON AN INVOICE: BALANCE DUE, BILL TO, ORDER NUMBER, UNIT PRICE
• PURPLE: FAMOUS GOATEES: BILLY GOAT, COLONEL SANDERS, DOCTOR STRANGE, THE DEVIL

• My rating: Hard
• My score: 1 mistake

The key to my “purple first” success today was saying aloud to myself “not by the hair on my chinny chin-chin”. This in turn made me think of Colonel Saunders and his resplendent white goatee and although I have never seen Doctor Strange I was pretty sure I’d seen a poster of Benedict Cumberbatch sporting some hair on his chinny chin chin.

The other groups I found much harder and I made a mistake assembling the green group.

I knew that the thread was something to do with buying tickets, but didn’t understand WILL CALL, so selected ORDER NUMBER.

In the process, I remembered a recent concert I went to where the only entry was via MOBILE TICKET, but my phone battery was at 1%; I made it in just as the screen died.

Yesterday's NYT Connections answers (Sunday, August 9, 2026, game #1155)

• YELLOW: TAKE A DIFFERENT APPROACH: CHANGE TACK, COURSE-CORRECT, RECALIBRATE, SHIFT GEARS
• GREEN: CLASSIC CHILDREN'S BOOKS: CAPS FOR SALE, CORDUROY, GOODNIGHT MOON, MADELINE
• BLUE: MOVIE SUBGENRES: BUDDY COP, NEO-NOIR, SPACE OPERA, SPAGHETTI WESTERN
• PURPLE: THINGS PEOPLE TRY TO GET OUT OF: BAD DREAM, ESCAPE ROOM, JURY DUTY, LABYRINTH

What is NYT Connections?

NYT Connections is one of several increasingly popular word games made by the New York Times. It challenges you to find groups of four items that share something in common, and each group has a different difficulty level: green is easy, yellow a little harder, blue often quite tough and purple usually very difficult.

On the plus side, you don't technically need to solve the final one, as you'll be able to answer that one by a process of elimination. What's more, you can make up to four mistakes, which gives you a little bit of breathing room.

It's a little more involved than something like Wordle, however, and there are plenty of opportunities for the game to trip you up with tricks. For instance, watch out for homophones and other word games that could disguise the answers.

It's playable for free via the NYT Games site on desktop or mobile.

Looking for a different day?

A new Quordle puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Sunday's puzzle instead then click here: Quordle hints and answers for Sunday, August 9 (game #1658).

Quordle was one of the original Wordle alternatives and is still going strong now more than 1,500 games later. It offers a genuine challenge, though, so read on if you need some Quordle hints today — or scroll down further for the answers.

Enjoy playing word games? You can also check out my NYT Connections today and NYT Strands today pages for hints and answers for those puzzles, while Marc's Wordle today column covers the original viral word game.

SPOILER WARNING: Information about Quordle today is below, so don't read on if you don't want to know the answers.

Quordle today (game #1659) — hint #1 — VowelsHow many different vowels are in Quordle today?

• The number of different vowels in Quordle today is 4*.

* Note that by vowel we mean the five standard vowels (A, E, I, O, U), not Y (which is sometimes counted as a vowel too).

Quordle today (game #1659) — hint #2 — repeated lettersDo any of today's Quordle answers contain repeated letters?

• The number of Quordle answers containing a repeated letter today is 1.

Quordle today (game #1659) — hint #3 — uncommon lettersDo the letters Q, Z, X or J appear in Quordle today?

• No. None of Q, Z, X or J appear among today's Quordle answers.

Quordle today (game #1659) — hint #4 — starting letters (1)Do any of today's Quordle puzzles start with the same letter?

• The number of today's Quordle answers starting with the same letter is 0.

If you just want to know the answers at this stage, simply scroll down. If you're not ready yet then here's one more clue to make things a lot easier:

Quordle today (game #1659) — hint #5 — starting letters (2)What letters do today's Quordle answers start with?

• G

• C

• S

• E

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

Quordle today (game #1659) — the answers

(Image credit: Merriam-Webster)

The answers to today's Quordle, game #1659, are…

• GAVEL
• CRUSH
• SPEED
• ETHOS

I’m glad I held back from getting GAVEL until later in the game.

Before eliminating the letters there were dozens of possibilities, but as the letters vanished it reduced down to one.

Daily Sequence today (game #1659) — the answers

(Image credit: Merriam-Webster)

The answers to today's Quordle Daily Sequence, game #1659, are…

• PRISM
• SNAKE
• CRAMP
• SALON

Quordle answers: The past 20

• Quordle #1658, Sunday, 9 August: KAPPA, WIDEN, FOLIO, STEEP
• Quordle #1657, Saturday, 8 August: FUROR, BUGLE, REUSE, SCOOP
• Quordle #1656, Friday, 7 August: AMISS, SHIRK, SALSA, RABBI
• Quordle #1655, Thursday, 6 August: AMUSE, BRAND, CEDAR, BLURB
• Quordle #1654, Wednesday, 5 August: MOUND, WRONG, STOOD, WEIRD
• Quordle #1653, Tuesday, 4 August: GOOEY, TARDY, HUMAN, CHALK
• Quordle #1652, Monday, 3 August: FUNGI, MUDDY, DEBUT, MANIA
• Quordle #1651, Sunday, 2 August: BOOTH, EASEL, REACH, MAKER
• Quordle #1650, Saturday, 1 August: OMEGA, BIRCH, SMOCK, EERIE
• Quordle #1649, Friday, 31 July: MANIA, SINGE, STOOL, LOFTY
• Quordle #1648, Thursday, 30 July: MODAL, IDLER, CHUMP, LUMPY
• Quordle #1647, Wednesday, 29 July: GAWKY, ALLOY, AUDIT, LATCH
• Quordle #1646, Tuesday, 28 July: SMITE, ROACH, RUDDY, GLOBE
• Quordle #1645, Monday, 27 July: TAFFY, FIGHT, GUILD, WAGER
• Quordle #1644, Sunday, 26 July: STOMP, LIMIT, FUNKY, STEAD
• Quordle #1643, Saturday, 25 July: SALON, SHEEN, BURST, GOURD
• Quordle #1642, Friday, 24 July: DOLLY, EDIFY, LAGER, PRANK
• Quordle #1641, Thursday, 23 July: LADEN, EVICT, WOVEN, SHIRE
• Quordle #1640, Wednesday, 22 July: COYLY, SINGE, AWASH, AWOKE
• Quordle #1639, Tuesday, 21 July: TRUSS, AXIAL, SWINE, THING

• The system connects directly to a standard alternating-current power outlet
• Sulfur co-implantation gave engineers precise control over individual diamond qubits
• Single-qubit fidelity reportedly reached 99.98%

German startup Saxon Q has launched the world's first portable diamond-powered quantum computer operating at room temperature without requiring complex cryogenic cooling equipment.

The system fits inside a standard server rack, connects directly to a conventional alternating-current power outlet, and supports configurations reaching 128 qubits.

The company says larger 512-qubit systems will become available next year, while its long-term roadmap aims to exceed 10,000 qubits after 2030.

Synthetic diamonds replace cryogenic cooling with a different quantum approach

The system relies on nitrogen-vacancy defects created inside synthetic diamonds, where nitrogen atoms occupy positions beside empty spaces within the crystal structure to function as quantum bits.

Scientists manipulate these qubits using lasers and microwave pulses, allowing them to represent quantum states extending beyond conventional binary computing methods.

Although nitrogen-vacancy technology has existed for years, producing stable systems beyond 10 qubits has remained difficult because creating reliable defects at larger scales proved technically challenging.

Saxon Q says its breakthrough came from introducing sulfur alongside nitrogen during manufacturing, allowing engineers to improve stability while gaining greater control over individual qubits.

“That's the key technology point. Because the sulfur lifts the chemical potential to a point that it's negatively charged,” said Marius Grundmann, professor of experimental physics at Leipzig University and co-founder of Saxon Q.

“The sulfur supplies the electron; the sulfur also makes the vacancy attached to the nitrogen with a very high yield.”

According to the company, this manufacturing process enabled qubits to achieve 99.92% fidelity before error correction, representing fewer than one error for every 1,000 operations.

However, it later updated its data, saying subsequent measurements increased single-qubit fidelity to 99.98%, although those figures have not yet received independent verification.

The hardware is currently available in rack-mounted systems supporting up to 128 qubits, while 512-qubit configurations are expected next year and expansion beyond 10,000 qubits remains the long-term objective after 2030.

Practical deployment offers advantages while scalability questions remain unresolved

Unlike many quantum systems that require cryogenic refrigeration and extensive infrastructure, Saxon Q says its hardware can operate directly within conventional computing environments without dedicated cooling facilities.

“We have a fully functioning quantum computer… We have a quantum computer that can execute quantum code that you can reach via the network, and it is a multiuser, multitasking, multicore system,” Grundmann added.

The company believes local deployment could reduce dependence on cloud-based quantum services for applications including robotics and autonomous driving, where communication delays may become unacceptable.

Researchers still caution that comparing nitrogen-vacancy systems directly with superconducting quantum computers remains difficult because both technologies emphasize different engineering approaches and operating characteristics.

Existing studies generally indicate superconducting systems deliver faster processing, although the balance between processing speed, cloud latency, and practical deployment remains uncertain across different workloads.

Another unresolved challenge involves increasing chip density, since current devices support only eight or 16 qubits per chip.

Future applications may require hundreds of thousands or even millions of interconnected qubits operating within significantly larger quantum arrays.

Via Live Science