Sony spent years turning over-ear canceling into something people actually wear all day, and the WH-1000XM5, priced at $199.99 (was $400), remains the pair that made that habit stick. Eight microphones and a pair of processors hush cabin rumble, subway grind, and open-office chatter so thoroughly that music and calls sit on a darker background than most rivals managed when these launched a few years ago.

Eight listening microphones fitted on the headphones monitor both what happens outside and inside the cups. Combining them with the V1 processor and a QN1 noise chip allows the system to automatically fine-tune noise cancelation for glasses, hair, leaks around the cups, and changes in cabin pressure. First to vanish are the low hum of airline engines and the persistent drone of your office’s HVAC system. Then mid-range office chatter begins to dissipate.

Sale

Sony WH-1000XM5 Premium Noise Cancelling Wireless Headphones, Black

• PREMIUM NOISE CANCELLATION: Two processors control 8 microphones for unprecedented noise cancellation. With Auto NC Optimizer, noise canceling is…
• MAGNIFICENT SOUND: Engineered to perfection with the new Integrated Processor V1.
• CRYSTAL CLEAR HANDS-FREE CALLING: 4 beamforming microphones, precise voice pickup, and advanced audio signal processing.

Even with the noise cancelation turned on, the battery life is 30 hours, and 40 hours when it is turned off. A USB-PD charger will provide three hours of listening for every three minutes of charging, and a full USB-C charge should take roughly three and a half hours, however real-world testers have reported matching or even exceeding those figures when using them at a moderate level.

One thing that has clearly improved is the feel, as soft fit leather pads and a smaller headband reduce clamp pressure, allowing you to wear them for hours without feeling like they’re squeezing your head. The cups now swivel to lie flat, but they no longer fold up into a little ball like they used to, so Sony had to tough up the shell to protect them. If you’ve ever jammed a pair of XM4s into a side pocket of a bag, you’ll understand what we mean.

The sound originates from new 30mm carbon fiber drivers that have been adjusted to be warmer and more even than the previous ones. The bass is still under control, the mids are clean, and if you have an Android phone that supports LDAC, you can get even more resolution out of your music. They also have a system called DSEE Extreme, which aims to restore the quality of compressed streams. The headphones include a full equalizer, 360 Reality Audio support, Adaptive Sound Control, which switches modes automatically based on what you’re doing, Speak-to-Chat, which pauses your music when you start talking, and Quick Attention, which allows you to cover a cup and hear what’s going on in the room.

Long-time testers believe they have made significant improvements in terms of call quality. Four dedicated voice mics, paired with improved processing, do an excellent job of preserving your voice’s natural tone while also eliminating background keyboard clatter and street noise. You may now connect two devices at once, and the touch swipes on the right cup handle allow you to adjust volume, tracks, and calls. A tactile button on the left cup activates and deactivates the noise cancellation and ambient settings.

An MIT committee report says AI can produce credible responses to almost any written assignment in its undergraduate curriculum and has driven measurable shifts in campus culture in under three years. In the EU, systems that monitor students during tests are high-risk under the AI Act and emotion recognition in education has been prohibited since February 2025.

MIT says artificial intelligence can now credibly complete almost any undergraduate assignment it sets. Its ad hoc AI committee published the finding this week, Futurism reported.

The list is not narrow. The report covers essays, maths and science problems, proofs and coding assignments, and says the models produce credible solutions across all of them.

The more interesting finding is not about cheating. In under three years the technology has driven what the committee describes as major shifts in campus culture.

Attendance at office hours is down. Participation in online discussions has fallen, and the report cites anecdotal evidence of fewer study groups in dorms and libraries. None of that is a cheating problem.

Other universities have already reacted. Chicago’s law school banned phones and laptops in first-year classes, and Princeton dropped an honour code more than a century old.

The direction of travel is supervision. If you cannot trust the work, you watch the person doing it, which is a very old answer to a new question.

That is where European law starts rather than ends. On this continent the watching is the regulated activity, not the cheating.

Universities were already using AI to catch cheating before any of this, and TNW called the practice creepy at the time.

The AI Act has a view on it. Education is one of the areas it singles out for the strictest treatment short of an outright prohibition.

Systems used for monitoring and detecting prohibited behaviour of students during tests are high-risk under Annex III, alongside admissions and the evaluation of learning outcomes.

Those obligations were due this month. The Digital Omnibus pushed them back to 2 December 2027, which buys institutions and their software suppliers another sixteen months.

The prohibitions were not pushed. Emotion recognition in education has been banned outright since February 2025, and the EU can inspect models and fine providers.

So the American argument is about whether to watch students at all. The European one was settled first, and it is about how closely.

There is a clear repeating trend in agent deployments: The gateway is the first control teams reach for, but it is the one they are least ready to run. This is because gateways sit on top of identity and attribution layers that are mostly not there.

The first layer of risk is not hypothetical. In June, CISA added a LiteLLM flaw to its Known Exploited Vulnerabilities catalog after attackers were caught abusing it in the wild. The bug ran commands on the host through the gateway itself, and chained with a second flaw it required no credentials. It was one of seven common vulnerabilities and exposures (CVEs) disclosed in that single AI gateway in a month. This is the layer many enterprises reach for first to secure their AI agents.

When considering secure agent architecture, gateway controls should not be the first control. They should be the fifth.

Most models on the maturity of agent security describe the controls a company will need in the future. They tend to miss, from my experience, the more difficult problem of describing the brownfield scenario: In what order should these controls be layered in conjunction with an identity and access management system that is already in place?

If the control plane is unaware of which agent is acting, who delegated the work, what task the agent is to perform, and what credentials are being used, then the context is incomplete. A gateway may block clear policy violations, but will struggle to distinguish a justified action from one that is technically permissible but operationally inappropriate.

The pattern of failure is clear when sequencing these controls for agent production deployments: Enforcement is taken early, while the identity and attribution context it depends on has yet to be developed. Agent security functions as a dependency chain, with each control depending on context generated upstream.

The wrong starting point

Think about routing agent traffic via a new runtime gateway. A finance-reconciliation agent tries to alter a record in production. The gateway authenticates the user token and checks the API call. What it can’t observe is that the request is agent-initiated, that the agent is executing a more limited function, or that the request is part of a tool chain invoked by an untrusted artifact.

The credential is valid. The API call is permissible. The action contradicts the purpose of the delegation. The gateway is there, but its set of supports seems absent, so a costly control is applied to a very small part of the whole picture.

Limiting an agent’s privileges to those of the human principal is useful so the agent does not exceed the person it serves. However, having a privilege ceiling does not create separate attribution. Twenty agents might operate under a single person’s permissions and still need unique identities, audit logs, behavior profiles, and revocation paths.

Dependency-gated deployment

I call this process dependency-gated deployment. Upstream exit tests must be satisfied before any downstream control is considered operationally complete. Concurrent development of downstream controls is permissible.

Here are the six gates, and the proof that they work:

The six dependency gates for the agent security controls. Each control is contextualized by the gates above it. From the author’s analysis of production agent deployments.

Start with the agents you can actually name

To begin, recognize the production agents in open-source frameworks, cloud offerings, SaaS services, and developer tools. For each, record the owner, responsibility, lifecycle stage, allowed tools, data domains, and sources of credentials.

Bypass this step, and the organization will lose the first hour of incident response while they figure out what should have been obvious. The inventory identifies the asset that every control thereafter governs.

An agent needs its own identity, but it cannot lose the human behind it

An agent should not be buried in a developer token, a shared service account, or a human session. Simply knowing the caller is an agent is not sufficient. The control plane requires additional delegation context: Who delegated the work, what specific task the agent was instructed to execute, and which resources the agent needs the authority to access. Identity specifies which actor placed the call. Delegation is the answer to whose authority it acts, and for what reason.

Once that connection is cut off, the downstream logs attribute the reconciliation agent to the employee whose token it borrowed, and every action it takes is attributed to someone who did not start it.

Shrink authority before you inspect behavior

Once an agent can be identified, capabilities should be limited. Access restrictions should be time-bound to the task and limited to the tools and resources required to perform the task. This can be implemented using identity access management (IAM) features such as workload identity, token exchange, conditional access, and time-bound entitlements which the organization already possesses.

With regard to the 2026 Teleport study involving 205 security leaders, the access scope surpasses the predictive capacity of industry, maturity, or self-assurance concerning predicting AI-related incidents. For example, organizations with over-privileged AI reported a 76% incident rate, whereas AI incidents occurred in 17% of organizations under the least privilege. This indicates that access scope in the dependency chain is more important than context-aware runtime enforcement.

The primary principle is monotonic delegation. Every transfer of responsibility must preserve or diminish authority; under no circumstances should it increase authority. For the reconciliation agent, this means an agent who can view one ledger as opposed to one who inherits the employee’s access to all systems the employee can access.

Fix attribution before automating enforcement

Most audit stacks can capture what resource was accessed and which credential allowed access. In the agent deployments I have reviewed, this is the most commonly missed gate. Prior to utilizing an adaptive runtime policy, link any relevant tool invocation to the agent identity, initiating principal, task id, parent action, and outcome. After doing so, examine the telemetry: For one completed task, see if you can track down the initiator, the agent who executed it, the authority under which the action was taken, the tools utilized, and the outcome. In regulated environments, oversight that is not attributed cannot be justified.

Now the gateway earns its keep

The gateway can use registered identities, explicit delegation, scoped credentials, and attributable telemetry to question if this agent is authorized to perform this action, for this principal, within this task, involving this resource. Although the user’s credentials may provision write access to the finance-reconciliation agent, the gateway has situational context and so determines that it is out of scope. This is control’s point of greatest value. The most stringent controls should be applied at irreversible boundaries — payments, access policy changes, deletions, modifications of the production environment, and data exports.

Detection and the kill path come last

Behavioral baselines are developed last because distinguishable and attributable agent activity must be established to set a standard. Then, security teams are able to identify anomalous patterns of tool usage, unexpected cross-domain access, and deviations from their assigned tasks. Containment is more than just the disabling of a single directory object: A proper kill path entails disabling the agent’s identity, invalidation of active and derived credentials, blockage of tool activation, termination of active tasks, and isolation of the workload that contains the agent.

Start without replacing your IAM

Designing a whole new identity program is unnecessary. If the existing identity provider doesn’t treat agents as native object types, begin with an authoritative registry linked to the existing workload identities. Following this, extend agent and task identifiers as trusted execution contexts, implement short-lived credentials to mitigate inherited privileges, and include those identifiers in tool-call logs for subsequent gateway ingestion. The dependency model remains unchanged as vendor support matures.

Control gaps are measurable. In Okta’s 2026 survey, only 34% of executives said their organization always applies the same level of security rigor to its agentic workforce as to its human workforce. The last control from the chain cannot be applied first to close that gap.

What to do in the next 30 days

Begin with 10 production agents. For each one, identify the owner, purpose, approved tools, and credentials. By now, you should have the beginnings of an agent registry and perhaps your first insights on governance.

Test attribution. Find out if IAM and logging can tell each agent apart from the human or service that delegated the task. If this kind of differentiation is not possible, a gateway would be operating without any visibility.

Reconstruct one completed agent task within an action chain, from start to finish, including downstream effects. Wherever the chain breaks is where your deployment falls short.

Adding downstream enforcement ahead of required context breaks agent security. Maturity models describe the destination. A build order gets you there without breaking production along the way.

Nik Kale is a principal engineer specializing in enterprise AI platforms and security.

Welcome to the VentureBeat community!

Our guest posting program is where technical experts share insights and provide neutral, non-vested deep dives on AI, data infrastructure, cybersecurity and other cutting-edge technologies shaping the future of enterprise.

Read more from our guest post program — and check out our guidelines if you’re interested in contributing an article of your own!

• AWS plans to add 2 million Nvidia GPUs between 2027 and 2028
• The expanded deployment follows an earlier commitment exceeding one million chips
• New Vera-based CPUs will support increasingly demanding artificial intelligence workloads

Amazon Web Services and Nvidia have expanded their long-running partnership to add far more computing power for artificial intelligence workloads.

The cloud provider intends to add two million more GPUs across its global infrastructure between 2027 and 2028.

Cars in European cities are smaller than ever. But as Europe’s appetite for microcars has grown, the cute Italian ‘yoghurt pots’ have largely given way to small Chinese EVs. Even Smart, the the iconic ultracompact car brand, has moved manufacturing to China.

Spanish startup Liux thinks it can compete in a crowded market with a tiny electric car built around sustainability.

Following the tracks of the Microlino out of Switzerland, Liux is trying to carve out a place in the market with its upcoming microcar, the Liux Big. The “Big” name is a joke. It is small enough to park at a right angle to the curb; but the name also reflects the oversized ambitions of a team that rarely takes the expected route.

“The idea of ​​a European car does not exist,” Liux co-founder Antonio Espinosa de los Monteros told TechCrunch. Coming from the CEO of a startup whose cars are already making headlines for being “made in Spain,” it was a surprising take — and one that reveals a lot about the company’s priorities.

Liux’s batteries are not made in Europe, but they are rechargeable at home, including with power generated from solar panels. The car is also meant to be easy to maintain and avoid the faster obsolescence cycles of modern cars. Perhaps most notably, its fiber body is made from a novel linen-based biocomposite designed so the material can later be extracted and recycled. 

“One thing that’s very clear for David and me is that recycling isn’t just a lab concept. You can recycle almost anything in a lab. What makes something recyclable has to do with how it’s built,” Espinosa said. “When you build something, you have to try to preserve the integrity of the material and the components so that a second life is possible.”

“Real circularity” is where Espinosa comes from; he previously cofounded Auara, a Spanish B Corp selling natural mineral water in bottles that are both recycled and recyclable. But after a larger player acquired this successful brand, he embarked on a new chapter with Sancho as his co-driver.

When it comes to cars, Sancho is in the driver’s seat, with low emissions on the radar. His specialty is engineering electric vehicles that can rival gas-powered ones. Before Liux, his most impressive feat was the Bóreas, a hybrid supercar unveiled at the 24 Hours of Le Mans in 2017. But after a fallout with his former partners, he and Espinosa teamed up to found Liux.

Liux’s first prototype, the Animal, combined their expertise: The fully electric five-seater was made almost entirely from recycled or plant-based materials. And yet, the two co-founders decided to pivot soon after unveiling the SUV to the world in 2022. It was then that they determined that their odds of completing homologation would be much higher with a smaller car.

Fast forward to 2026, and Liux has secured Europe-wide homologation for the Liux Big, which it expects to start selling in the first half of next year. In the meantime, the company has grown to 65 employees and is getting ready to ramp up production across three facilities in Spain.

These include the plant that TechCrunch visited in Azuqueca de Henares, about a one-hour drive from central Madrid.

Liux’s Azuqueca factory is small because it follows Toyota’s “lean management” principles and performs only the last steps of the process, said its head of production Beatriz Belda González, a Spanish-born engineer who previously worked for BMW in Munich. But don’t let its size fool you: Liux says its production capacity could reach 20,000 cars a year by 2030.

It is still too early to gauge demand, but more than 7,500 people have joined the waiting list for a Liux Big. Joining doesn’t require a fee, but the list has helped the startup learn more about its prospective buyers. The most represented profile is a 55- to 60-year-old city dweller, and Liux is now assuming that the Liux Big will often be a household’s second car.

This may dampen hopes that microcars could challenge traditional car ownership, but Liux has to pick its battles, Espinosa said. Rather than trying to guess where the market is going, or how fast, the startup is keeping the door open to partnerships with companies that manage B2B fleets and others that could help make its cars autonomous.

For Espinosa, the Liux Big could already make a difference by offering a more sustainable option that is also affordable. The startup hasn’t confirmed its final price tag, but said it will be below €18,000 — about $21,000 — before potential EV subsidies. This puts it at the higher end of the price range for microcars, but Liux hopes it will punch above its weight — literally. 

According to Liux’s head of R&D, Celso Fernández Llorens, weight and size limitations are a huge constraint in this category. In his view, most microcars are fairly similar, despite the fact that European authorities differentiate ultralight L6e four-wheelers from slightly heavier L7e ones. Liux, however worked around these constraints to make the most of its L7e homologation.

Thanks to a litany of decisions large and small, the startup managed to fit a 260-liter trunk into the car. But most of its efforts were geared toward making sure users feel like they are driving a car, rather than a two-wheeler. That’s also closely tied to safety, Sancho said: you don’t want a car that’s only lightweight because its frame can’t withstand a crash, or that will topple on the first turn.

With this in mind, and despite the fact that its category doesn’t even require crash tests, Liux has been testing and showcasing the Liux Big’s ability to slalom, brake, and perform other maneuvers. The startup demonstrated some of those capabilities to TechCrunch during a short ride and test drive in its upcoming off-road version.

For now, its main model will have two versions: 15 kWh and 20 kWh. A cargo version is also planned, and with Sancho on the team, the temptation to build a supercar is never far away. In a LinkedIn post, the company noted that it doesn’t intend to be “a one-car brand.”

First, though, Liux will use the €16 million it has secured so far (about $18.5 million, including European funding) to bring the Liux Big’s urban version to market through partnerships with car dealerships across Europe.

The showroom where we met is also a preview of Liux’s future sales experience, head of brand Ana Terrado Leyva said. She pointed to textile screens, 3D models showcasing the Liux Big’s three color options — two more than the Ford T — and a linoleum floor as a nod to linen. These aesthetic choices, she said, are another way Liux hopes to stand out from its Chinese competitors.

Maybe the idea of ​​a European car does exist.

There are several things you should check.

Don’t blame Android Auto first — Check the music app

Spotify

Use wired and wireless connections to narrow it down

Alex Photo Stock/Shutterstock

AI AND ML

Disclosure optional, quality mandatory

The Debian community has voted to allow members to use generative AI when creating their contributions, with the caveat that developers remain responsible for code quality.

The Linux distro’s community recently decided to develop a policy on use of AI-assisted coding tools, and asked participants to vote on one of eight proposals that included an outright ban, cautious use, or just avoiding LLMs because of their impact on Earth’s environment.

The rules of the vote saw community members asked to rank each of the eight proposals. Just under 600 people voted, but Debian’s election team rejected many for unspecified reasons, leaving almost 450 valid votes to count.

When the tallying was done, proposal E – “Responsible Use of Generative AI” – won the day.

The proposal means “Debian neither endorses nor prohibits the use of generative AI tools in the development, maintenance, or documentation of software, packaging, documentation, and other media published within the Debian Project.”

The proposal advocates that approach because “such tools can substantially improve the productivity of contributors when used responsibly, allowing volunteers to spend more of their limited time on work that requires technical expertise, judgment, review, and collaboration.”

The proposal also spells out that the Debian community doesn’t believe “AI made a mistake” is an excuse for sloppy contributions.

“The Debian Project nevertheless expects that all contributions submitted to Debian, regardless of how and with which tools they were produced, satisfy the same standards of quality, correctness, maintainability, and legal compliance,” the proposal states. “The use of a generative AI tool does not diminish the contributor’s responsibility for the work they submit. Contributors are expected to understand, review, test, and, where appropriate, modify AI-assisted output before incorporating it into Debian.”

Woe betide the dev who doesn’t check their AI-generated code, as the proposal also states, “Blindly accepting or uploading AI-generated material without appropriate human review is inconsistent with Debian’s established development practices.”

Another element of the proposal tries to make acknowledging the use of AI a norm.

“We encourage our contributors to disclose whether a contribution was made with AI assistance, but do not require them to do so,” the text states.

As our FOSS aficionado Liam Proven wrote in his report on the vote, the Gentoo Linux team has banned use of AI, while NetBSD and OpenBSD don’t want any clanker-written code contributions.

Linus Torvalds, arguably the most influential figure in the FOSS community, welcomes AI-generated contributions.

In July he declared “Linux is not one of those anti-AI projects” – and recently proved it by using AI to squash a tricky bug. Torvalds has also sometimes complained about AI, such as his May observation that AI-generated bug reports sometimes made the Linux security mailing list “unmanageable.” ®

Bill Gates has published a memo proposing a tax on AI, a category of jobs reserved for humans and new institutions to manage the transition, and told GeekWire that industry leaders are worried privately while staying quiet publicly. He proposed a robot tax in February 2017, the same month the European Parliament adopted robotics rules that left the tax out.

Bill Gates says the people building AI are worried in private and quiet in public. “I know they’re all worried. Or all of them that I know, which is basically everybody but Elon,” he told GeekWire.

The reason he is talking is a memo he published this week. It argues the industry is crossing the safety lines it set for itself, and that nobody is preparing for what follows.

TNW covered the proposals when it landed. A tax on AI, a category of jobs reserved for humans, and new institutions to manage the transition.

The interview adds the part he did not put in writing. “Even lack of control that I thought would be many years from now, we’re seeing lack of control,” he said.

On the tax, Gates is repeating himself. He first proposed taxing robots in February 2017, when the argument was about factory automation rather than models, and the reasoning was the same.

Europe answered that question the same month. The European Parliament voted 396 to 123, with 85 abstentions, on rules for robotics, and the tax was not in what passed.

What passed instead is worth reading. The resolution asked the Commission to look at the viability of member states’ social security systems as robots spread, which is the diagnosis behind the remedy MEPs had just dropped.

It also floated something stranger. The adopted text raised electronic personhood for the most sophisticated autonomous robots, an idea that has gone nowhere since.

The rapporteur was not pleased. Mady Delvaux said certain coalitions in the Parliament had rejected a forward-looking debate about workers, and the robotics industry welcomed the outcome.

Nine years on the position has not moved. The AI Act regulates risk and requires human oversight, and TNW has asked whether it can protect jobs without stifling innovation.

It does neither of the things Gates wants. It does not tax a model, it does not reserve a category of work for people, and no member state has proposed either.

The evidence base is thinner than either side admits. TNW has argued that what AI is doing to jobs in Europe looks different from the headline numbers.

Which leaves the memo where it started. A man who proposed this in 2017 is proposing it again, and Europe has already voted.