from the the-future-will-look-a-lot-like-the-past dept

After burning $80 billion on the failed metaverse pivot — and billions more on desperate “me too” offerings in the AI space — Meta is looking to leverage its massive ad dominance to colonize the agentic AI (or personal assistant) market. So they recently introduced Muse, an AI agent represented by cutesy and personalized avatars, capable of doing semi-complex tasks like booking appointments and shopping.

The capabilities and novelty of the agent, as usual, tend to overshadow the potential privacy threats of giving unethical companies access to even more personal accounts and data. And right on cue, security researchers found a massive zero-day flaw in Muse that allowed any app or terminal command to gain access to the authentication token linking users to their Muse account. 

That not only created potential access to the Muse account (and everything it had access to in turn), it allowed for user surveillance that wasn’t transparent to the Muse user. Great stuff!

Meta marketing had spent a lot of time claiming they’d kept security and privacy at the forefront of Muse’s design so it was a bit of an embarrassing launch. And it certainly didn’t assuage fears about giving big companies like Meta access to even more personal information and login data:

“To me, the bar is infinitely higher in terms of the security of these apps. They don’t have to be perfect, but when you take a look at Muse, it’s like they didn’t, in my opinion, think about security, which is really worrisome,” Wardle said. “At the very least, they should be thinking about security from the very start, and they are just not.”

Meta’s adventure in mass-appeal agentic AI also ran into a roadblock when Amazon announced it was banning the agent from shopping on Amazon. Amazon’s announcement, made about 12-hours before Wardle discovered the vulnerability, claimed Muse was an “unauthorized AI agent” that “violates Amazon’s Conditions of Use:”

“We think it’s fairly straightforward that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service provider decisions about whether or not to participate. This helps ensure a safe, secure, and reliable customer experience, and it is how others operate including food delivery apps and the restaurants they take orders for, delivery services apps and the stores they shop from, and online travel agencies and the airlines they book tickets with for customers. Agentic third-party applications such as Muse have the same obligations, and we’ve requested that Meta remove Amazon from the experience.”

Amazon had already sued Perplexity over its comet browser (though unsuccessfully so far), and taken steps to block shopping agents from both Google and OpenAI. You can see how this could steadily devolve into an annoying walled-garden arms race that erodes the functionality of everybody’s agents.

There’s obviously interesting potential in agentic AI (aka software), but like so many other arenas, our failure to enforce antitrust law, and obsession with lobotomizing our regulators, means the market is likely to be dominated by the biggest companies. Companies that already spent the last decade making it clear they have very little ethics. And are keen to anti-competitively ratfuck their way to market domination.

Which is to say that Amazon’s blockade of Muse is likely only the beginning. In that sense, the future is going to, in a lot of ways, look very much like the past. It’s also another reason why you’d like to see the sector disrupted by cheaper, on-device, open source, open weighted AI alternatives where ideally the end user gets more transparency, better potential security, and more control.

Instead, I suspect 2027 is going to be heavily dominated by companies like Meta (and Amazon) scaring the government about China, resulting in terrible, protectionist, new AI legislation ghost written by its own lawyers.

Filed Under: agentic, ai, anticompetitive, bots, competition, muse, privacy, security, shopping

Companies: amazon, meta

We’ve yet to find any definitive evidence that there’s biological life on present-day Mars, but to say it’s a dead planet isn’t exactly accurate. Since the first Viking lander touched down in 1976, a revolving cast of humanity’s robotic envoys have worked on and around the Red Planet — and as access to space becomes cheaper and more routine, the mechatronic population of Mars will continue to grow.

Given the number of landers, rovers, and orbiting spacecraft that have been sent to study Mars over the last 50 years, you might be surprised to find that the communications systems in place to transmit all that critical scientific data back to Earth aren’t nearly as robust as you’d think. While it’s understandable that the first craft to arrive at Mars had to operate in isolation, even the flagship Perseverance and Curiosity rovers carry their own high-gain radio systems so they can communicate directly with Earth. Given the incredible premium put on the mass of an interplanetary craft, each mission that needs to bring along its own link back to Earth effectively reduces its payload of much scientific equipment.

It’s not that satellites in orbit around the planet aren’t used to relay signals between Martian ground assets and their controllers back on Earth. In fact these relay links are used extensively for bandwidth-intensive tasks such as image transfers. But it’s also true that the craft currently available to act as intermediaries between the two planets aren’t terribly well suited to the task. The current fleet of Mars orbiters were conceived primary as research vehicles, and so every decision regarding their design and positioning around the planet was made with that goal in mind. What relatively limited capability they do have as communication relays is further hindered by the age of their hardware.

But after decades of false starts and shifting budgets, NASA is closer than ever to finally establishing the Mars Telecommunications Network, a dedicated high-bandwidth communication relay that will ensure current and future missions always have a way to phone home.

The Fragile Mars Relay Network

Mars Odyssey, Mars Express, the ExoMars Trace Gas Orbiter, and the Mars Reconnaissance Orbiter currently make up what’s known as the Mars Relay Network (MRN). The Mars Atmosphere and Volatile Evolution (MAVEN) spacecraft was also part of the MRN, but communication was lost with the vehicle in December of 2025 and as of this summer NASA officially declared the mission over.

The MRN has been absolutely invaluable for many high profile missions, and there’s an excellent chance nearly every picture you’ve ever seen of the Martian surface was at one point routed through it on its way back to Earth. But for the spacecraft that comprise the MRN, shuttling bits across the black is more of a part-time job. Their individual primary missions are scientific in nature, and it shows.

Consider the Mars Express. Launched by the European Space Agency in 2003, it’s primary mission goals included releasing a small rover, analyzing the Martian atmosphere, and studying the planet’s closest moon, Phobos. To accomplish these tasks and to reduce the amount of propellant needed, the ESA put Mars Express into a highly elliptical orbit that sees the spacecraft swoop down to within 300 kilometers (185 miles) of the surface before heading back out to a distance of 10,000 km (6200 miles). Unfortunately, the rotation of Mars in combination with this exaggerated orbit means that rovers on the surface can’t maintain a consistent link to the vehicle, and as such its usefulness as a communications relay is limited to relatively narrow windows.

But in 2026, the real problem with the MRN is how old its constituent members are. MAVEN had already been flying for 12 years when it went offline, but that’s nothing compared to its peers. Mars Express has been on the job for 23 years, while Mars Odyssey has been in orbit for an incredible 25 years. Even the baby of the group, ESA’s ExoMars Trace Gas Orbiter, was launched over a decade ago.

Those are very impressive durations for missions of this type, so much so that Mars Express and Mars Odyssey are the two oldest operational spacecraft in orbit around a planet other than Earth. While their continued operation is a testament to the engineering that went into them, the Mars Relay Network has more than earned its retirement.

An Overdue Upgrade

Given their age, it’s not as if there’s any disagreement about whether or not the spacecraft of the MRN are due for a replacement. Similarly, few would deny the operational benefit of a dedicated communications relay in orbit over Mars. Unsurprisingly, discussions on both points have been going on since the early 2000s.

NASA’s original plan for what they called the Mars Telecommunications Orbiter (MTO) would have put the spacecraft in orbit around the Red Planet back in 2010. The design concept for the vehicle called for two X-band transmitters and two Ka-band radios, in addition to an experimental laser communication system that could return data to Earth at an unprecedented rate.

Unfortunately, the MTO concept was cancelled just a few years after its inception as NASA needed to allocate the money towards other missions such as the final servicing mission for Hubble and the then in development Mars Science Laboratory (MSL) mission which would go on to deliver Curiosity to the surface in 2012.

The general idea has been floating around ever since, being reintroduced in budget discussions every few years. Each proposal has failed to gain the necessary support, in part due to the vague nature of the mission. Usually when NASA or the ESA spends hundreds of millions of dollars to send a spacecraft to Mars, it has a long list of science objectives to justify the price tag. But a communications relay that carries few if any secondary science payloads is a harder sell.

Sample Return Consolation Prize

Given all the false starts the project has had over the last two decades, it’s somewhat ironic that the only reason the space agency is able to allocate significant funds for the mission now is because of the collapse of the Mars Sample Return (MSR) initiative. A budget reconciliation package passed by Congress in July of 2025 allocated $700 million to revive the Mars Telecommunications Orbiter program, but stipulated it had to be built by one of the commercial companies that had contributed towards the MSR mission, and that it should be ready to launch by 2028.

In May of this year NASA put out a request for proposals to build what it was now calling the Mars Telecommunications Orbiter (MTN), and earlier this month they awarded Blue Origin with a fixed-price contract to develop, build, launch, and operate the spacecraft.

As with previous incarnations of the MTO, Blue Origin’s spacecraft will provide a high-bandwidth link back to Earth which can be utilized by multiple current and future Mars missions. The vehicle itself will feature a number of ports which can mount scientific instruments, deployable sub-relays, or potentially even a lander. Technically the contract only requires that the orbiter be capable of carrying a 20 kilogram science payload, but presumably Blue Origin has expanded on that idea so they can offer Martian “rideshare” opportunities for other companies while still achieving NASA’s stated design requirements.

Given how quickly all of this is coming together, there’s still not a lot of public information on Blue Origin’s spacecraft. But  we can glean a few interesting technical details from NASA’s original proposal request, such as the requirement that the relay must offer at least one terabyte of non-volatile store-and-forward capability, and should be able to achieve a 150 Mbps link from Mars to Earth.

Interestingly, despite the impressive demonstration of the Artemis II Optical Communications System (O2O) earlier this year, NASA specifically noted that support for optical communication was not a requirement of the Mars Telecommunications Network contract — though it would consider the feature a bonus should it be included.

Not Everyone is Happy

With the funds guaranteed by federal mandate and a contractor already building hardware, it seems like the Mars Telecommunications Orbiter is now closer to reality than ever before. But of course, nothing is ever truly guaranteed when it comes to space. Production runs into snags, bugs are discovered in software, and occasionally rockets explode on the launchpad. In addition to this general uncertainty, the revived Mars Telecommunications Orbiter will also have to contend with politics.

Since the announcement that Blue Origin was selected for the MTN contract, Rocket Lab filed a formal protest with the Government Accountability Office. They claim that design presented by Blue Origin is “inconsistent with the eligibility criteria mandated by Congress” and that NASA made incorrect assertions about RocketLab’s own proposal.

Because Rocket Lab is protesting the decision, both NASA and Blue Origin must legally stop all work on the MTN project until the Government Accountability Office can review the selection process and make their determination. This process can last up to 100 days, meaning there might not be a ruling until the end of the year.

Under normal circumstances this might simply be a bureaucratic annoyance, but when it comes to a mission that needs to leave Earth during the periodic Mars launch windows that only come around every 26 months, a lengthy delay could ultimately push the project into the next decade.

If you hold a FedRAMP certification, the nearest deadline on your calendar is December 7, 2026. FedRAMP’s notice responding to CISA’s BOD 26-04 puts it plainly: “The Vulnerability Detection and Response rules will be mandatory for all cloud service offerings obtaining or maintaining FedRAMP Certification effective December 7, 2026,” with a grace period through March 7, 2027 for offerings operating under a corrective action plan.

It is also the deadline most programs have not fully scoped. The problem is not the date. It is that VDR and VER read like a scanning requirement and operate like something else entirely.

What the two rulesets actually change

Start with what is retired: the flat monthly-scan-and-POA&M model. Detection frequency is now set by certification class — under rule VDR-TFR-PSD, machine-based resources are scanned at least every 14 days at Class A, every 7 at Class B, every 3 at Class C, and at least once per day at Class D.

Machine verification and validation runs at least monthly for Rev5 holders, and as often as every three days at higher 20x classes.

Then the three provisions that reshape engineering work:

Remediation clocks are tiered and tight. Under VDR-TFR-PVR, fix deadlines are set by a vulnerability’s PAIN rating and its exploitability, running from 192 days at the low end to 12 hours at the extreme — a Class D offering with a PAIN-5 vulnerability that is both likely exploited and immediately remotely exploitable.

A 12-hour clock is not a ticket-queue SLA. It is a paging and ownership question, and it has to hold on a holiday weekend.

The burden of proof inverted. VER-EVA-AIA — “Assume It’s Automatable” — requires providers, in FedRAMP’s words, “to assume exploits are automatable by default, unless they have evidence providing otherwise.” Every deferral now needs a defensible artifact behind it, produced at volume, on the same clock as everything else.

Process failures count as vulnerabilities. Rule VDR-CSO-FAV states that providers “[MUST] treat problems or failures with their vulnerability detection and response processes as vulnerabilities.” If your detection pipeline silently stops, that is not an operational hiccup you fix quietly before anyone notices. The system that produces your evidence is itself in scope.

Read together, those change the deliverable. You are not being asked to scan more often. You are being asked to run a system that produces defensible, current, machine-readable answers about your own exposure — and to be accountable when it stops running.

December 7 is the first installment, not a one-off

The Consolidated Rules for 2026 reorganized FedRAMP into rulesets and started the clock on a much larger change.

Rev5 is not being maintained alongside 20x: FedRAMP describes it as “a legacy FedRAMP Certification process that is being replaced entirely by FedRAMP 20x,” and says providers “are expected to follow new rules and adopt new FedRAMP Practices from FedRAMP 20x into their FedRAMP Rev5 Certified cloud service offerings.”

The rules become mandatory for all stakeholders on January 1, 2027, and FedRAMP stops accepting new Rev5 applications on June 11, 2027.

So VDR and VER are not a detour you take before the real transition. They are the transition, arriving in installments — and that reframe is the most useful thing available to a program right now, because it changes sequencing.

Work scoped as “get through December” gets rebuilt in 2027. Work scoped as the first slice of continuous validation transfers.

What got removed tells you where this is going

Look at the structural changes rather than the deadline table. The System Security Plan and its appendices give way to a Certification Package Overview and a Security Decision Record. Plans of Action & Milestones, FedRAMP writes, “have been eliminated entirely and replaced with a list of Accepted Weaknesses.”

Continuous Monitoring becomes Ongoing Certification — renamed, FedRAMP explains, because “continuous monitoring” had “become synonymous with ‘vulnerability scans’” and the new requirements are “far broader than before.”

Every one of those was a place where the artifact stood in for the reality. FedRAMP was unusually direct about closing them, telling providers they will need to build or buy modern GRC capabilities and “populate them using automation based on real-world data where possible, rather than maintaining artisanal hand-crafted documents.”

Here is what deadline coverage keeps missing: almost none of this is a demand for new security. Access control, identity, encryption, logging, incident procedures, training — largely intact, largely reusable.

What changed is that describing them no longer counts as evidence of them.

Three judgments that separate the programs that make it

The work is still compliance; the deliverable is now engineering. What you hand over is a set of running validations that pull from the systems holding the truth — cloud configuration, identity provider, SIEM, CI/CD, ticketing — and emit machine-readable results on a schedule.

Providers must persistently validate their Key Security Indicators, of which CR26 currently lists 49 across ten categories. That is an entry requirement, which makes every transition plan an automation engineering plan underneath whatever it says on the cover.

Someone has to own “continuous.” Monthly monitoring had a due date, an owner, and a natural rhythm of catching up. A validation cadence has none of those. It runs, or it silently stops, and the difference is invisible until an assessor or a customer finds it.

Before building pipelines, answer the operational questions: who is paged when a validation fails, what the response time is, who notices when an evidence source quietly changes its API.

Design for the cadence, not the submission. FedRAMP defines persistently as “occurring in a firm, steady way that is repeated over a long period of time in spite of obstacles or difficulties” — a description of an operating state, not a date.

Teams that build toward a submission build a system tuned for a single moment and then rebuild it afterward.

The part that outlives FedRAMP

Once evidence is structured data rather than narrative, it stops belonging to a framework. The identity evidence satisfying a FedRAMP indicator is the same evidence a SOC 2 auditor wants and the same evidence a large customer’s diligence team asks for.

Compliance stops being parallel projects that each rebuild the same picture in a different vocabulary and becomes one substrate that many consumers read from.

The economics invert along with it. Point-in-time compliance costs rise with every framework and every region you add, because each addition is more description to produce and maintain. Continuous validation costs materially more to stand up and barely more to run.

December 7 is a hard date, and it deserves the attention it is getting. But financial-services supervisors, the EU’s resilience and product-security regimes, and enterprise procurement teams are converging on the same demand from different directions: show me current state, not last year’s description.

FedRAMP arrived first because it had the clearest mandate and the least patience. A team that builds this once has not solved a federal problem — it has built the capability every one of those demands will keep asking for.

anecdotes holds a FedRAMP 20x Class C certification, earned as a Phase Two pilot participant, using the anecdotes platform to run it. The same platform runs commercial compliance for more than 140 enterprise customers.

Standard basis: FedRAMP Consolidated Rules for 2026 and FedRAMP Notice NTC-0014. Rules and Key Security Indicators change through FedRAMP’s public rules process; confirm the live standard at fedramp.gov before baselining your plan.

Learn how Anecdotes helps you operationalize VDR & VER and download the technical solutions brief here.

Sponsored and written by Anecdotes.

The algorithms powering modern dating apps can be a maze. Finding a match, especially without a premium subscription, can be quite a hassle. The “swipe anxiety,” as they say, is a well-known and ugly reality.

Rivet takes a completely different approach to finding connections and romantic interests on a dating app. The dating network, launching today in the US, lets strangers play matchmaker for you.

How exactly does Rivet work?

The brainchild of former Tinder and Match Group executive Taru Kapoor, the app lets users act as both a dater and a matcher. “The pairs who get high Social Matching scores are introduced to each other, and each individual independently decides whether they want to match with someone and begin a conversation,” says the company. 

Rivet follows a “give-to-get” format, noting that it’s a win-win situation where users find love and also play an active role in helping other Ripeople come together. The app is free to download and is now available on Android and iOS.

However, the matchmaker concept isn’t unique to Rivet.

Which other dating apps have matchmakers?

One of the most well-known applications of its kind is Wingman. It lets friends and family members act as matchmakers by writing your profile bio and swiping through potentially interesting folks you should connect with based on their hobbies and interests.

In July this year, Hinge also launched a new feature called Friend’s Take, which lets friends and family members contribute their thoughts on your profile through text, voice, and video descriptions. 

Back in 2023, Tinder also introduced a tool called Tinder Matchmaker, which let your acquaintances access Tinder without even signing up, check profiles, and recommend them to you.

Rivet is reintroducing that concept by letting users act as matchmakers for others in the circle, or sign up as a user to find romantic interests for themselves.

Free legal streaming can cover a surprising amount of everyday viewing, but the trade-off is usually less control over advertising, title availability, offline viewing, and premium playback features. Paying does not automatically remove ads either, because many subscription services now offer lower-cost ad-supported plans alongside more expensive ad-free tiers.

The useful comparison is therefore not simply free versus paid. It is closer to a spectrum: free ad-supported streaming, paid ad-supported streaming, and paid premium streaming. What matters is which restrictions actually affect the way you watch.

Free and Paid Streaming Are No Longer Opposites

Several streaming business models now sit between completely free television and a premium subscription.

AVOD, or advertising-supported video on demand, lets you choose a movie or program and watch it with commercial breaks. FAST, or free ad-supported streaming television, usually combines advertising with scheduled channels that resemble traditional television. SVOD, or subscription video on demand, charges a recurring fee for access to a service, but that subscription can itself be either ad-supported or largely ad-free.

Tubi is a straightforward AVOD example. Its current advertising explanation says the service remains free because advertising funds it and that it does not currently offer an ad-free option.

Netflix demonstrates why the paid side is more complicated. Its current U.S. plan documentation lists Standard with ads alongside ad-free Standard and Premium plans. Paying for a streaming subscription and seeing commercials are therefore not mutually exclusive.

There are other models too. Library-supported services can provide access through participating public libraries, colleges, or universities rather than through advertising or a personal entertainment subscription. The practical differences among free legal streaming services therefore depend partly on how each service is funded and how its catalog is delivered.

If the distinction between scheduled free channels and free on-demand libraries matters to you, FAST and AVOD work differently even though both can be free and advertising-supported.

Free vs Paid Streaming at a Glance

The most useful comparison is not which category has the longest feature list. Compare the restrictions that affect your viewing: advertising, access to specific titles, offline use, playback quality, device limits, and regional availability.

No column wins every row. A person who mainly browses whatever looks interesting has different requirements from someone who follows one exclusive series, downloads shows before flights, or shares a subscription across several screens.

The First Thing You Give Up Is Control Over Interruptions

Advertising is the most visible cost of many free services. Instead of paying a recurring entertainment fee, you accept commercial breaks that help fund the service.

Tubi explicitly ties its free model to advertising and currently provides no ad-free Tubi tier. That makes the trade-off straightforward: access costs no subscription fee, but advertising is part of the viewing experience.

The same assumption becomes less reliable once you move into subscription streaming. Netflix’s ad-supported experience documentation says most movies and TV programs on eligible plans contain commercial breaks. It also notes that a small number of titles can be unavailable on the ad-supported experience because of licensing restrictions.

That means the useful question is not simply whether you pay. Ask what kind of advertising experience a particular plan includes. A cheaper subscription may reduce the financial difference between free and premium streaming without eliminating interruptions.

Live programming is another edge case. Netflix states that live events can contain commercial breaks across all of its plans, including ad-free experiences. If avoiding commercials is one of your main reasons for upgrading, verify the provider’s current rules for both on-demand and live content.

The Bigger Trade-Off Is Often What You Can Watch

Catalog access matters more than raw title counts. A free service can offer thousands of programs and still be a poor substitute if the one series, sporting event, or film you want is licensed somewhere else.

Free streaming should not automatically be dismissed as a warehouse of old programs. Nielsen’s March 2025 Gracenote FAST analysis reported that more than 70% of FAST programming in its analyzed dataset was produced since 2010.

The category has also continued to expand. Nielsen reported in August 2026 that global FAST channel count in its Data Hub had reached 2,172, with sports programming growing faster than the overall FAST channel count.

Those findings show that free streaming has developed beyond purely archival television. They do not mean a free catalog will contain the same titles as Netflix, Disney+, Max, Paramount+, or another subscription service.

Licensing determines where much of the difference appears. Rights can be sold for particular countries, platforms, time periods, or distribution models. A film might therefore be included with a subscription in one country, offered free with ads somewhere else, and unavailable on both services in another market.

Exclusive programming creates a stronger reason to pay. If a show or event is contractually tied to one subscription service, a large free catalog elsewhere does not replace it. Conversely, a free service may occasionally carry a title that is absent from the subscriptions you already have.

For that reason, paying for “more content” is too vague a justification. The better question is whether the subscription gives you reliable access to content you specifically value. When one exact film or series is the priority, checking where a title is legally streaming is more useful than comparing catalog-size claims.

Offline Viewing and Playback Flexibility Favor Some Paid Plans

Free streaming generally works best when you expect to watch while connected to the internet. Offline viewing is provider-specific, and paying does not guarantee that every download feature is unrestricted.

Netflix illustrates the distinction within one subscription service. Its current download-limit documentation says its ad-supported plans are limited to 15 total downloads per device per calendar month, with the limit resetting on the first day of each calendar month.

That is a useful reminder that “has downloads” and “has unrestricted downloads” are different claims. Providers can also impose title, device, or licensing restrictions on offline viewing.

Offline playback matters most when internet access is unreliable or expensive, such as during travel or on a commute. If you almost always watch at home on stable broadband, the feature may have little practical value even if a premium plan includes it.

Paying More Can Buy Better Quality and More Device Capacity

Some higher subscription tiers sell technical capability rather than simply a larger entertainment catalog.

Netflix’s current U.S. plan page lists 1080p playback and two simultaneous supported devices for both Standard tiers. Premium increases that to four simultaneous supported devices, adds 4K Ultra HD with HDR, allows downloads on more supported devices, and includes spatial audio. These are current Netflix plan differences, not universal features of paid streaming.

The hardware still has to support the feature. Paying for a 4K plan does not make a 1080p television display 4K. The title, television, streaming hardware, connection path, and service all need to support the required format.

If the television’s built-in software or hardware has become the limiting factor, comparing a Smart TV app, Google Cast workflow, and dedicated streaming stick can help separate a subscription limitation from a device limitation.

The same reasoning applies to simultaneous streams. A larger household may value extra concurrent devices, while someone watching alone gains little from paying solely for that allowance.

Free Streaming Can Have Significant Regional Restrictions

Streaming rights remain geographical whether a service is free or paid. Free does not mean globally available, and subscribing does not eliminate territorial licensing.

The Roku Channel is a clear current example. Roku’s official availability documentation says The Roku Channel is available in the United States, Canada, and the United Kingdom on Roku devices, web, and mobile, while access in Mexico is limited to Roku devices.

That creates several different kinds of availability to check. A service can operate in your country but lack a particular title there. Its mobile app may exist while its Smart TV app does not. A live sporting event can have different territorial rights from the service’s regular movie catalog.

These restrictions mean a comparison written for one country should not be treated as a universal catalog guide. Before subscribing mainly for one show or event, verify the title, plan, device, and country that apply to you.

What About Privacy and Ad Targeting?

Advertising introduces another consideration, but the price of a plan does not tell you by itself how private the service is.

Netflix explains that its ad-supported experience can use information such as viewing interactions and general location when selecting advertisements. It also says behavioral advertising can use activity from unaffiliated apps or websites unless the viewer has opted out where that option applies.

Info
iDo not assume that “free means your data is the product” or that paying automatically prevents tracking. Advertising controls and privacy practices vary by provider, plan, and region, so compare the policy of the service you actually use.

The presence or absence of a subscription fee is therefore not enough to establish a privacy hierarchy. If targeted advertising matters to you, check the provider’s advertising controls, privacy settings, account requirements, and regional privacy options rather than relying on labels such as “free” or “premium.”

A paid ad-supported plan can also be the middle ground. Deloitte’s March 2026 U.S. Digital Media Trends research reported that 68% of SVOD subscribers had at least one ad-supported tier, up from 46% in 2024. The finding applies to the U.S. survey population rather than streaming subscribers globally.

Before paying more to remove ads, compare whether an ad-supported streaming plan is worth the lower price based on how often you watch, the ad restrictions, catalog access, and the premium features you would actually gain by upgrading.

Bottom Line

Free streaming no longer means settling for a tiny library of obsolete television. Legal free services now cover substantial on-demand and channel-based viewing, but they commonly ask you to accept advertising, changing catalogs, regional restrictions, and fewer premium controls.

The strongest reason to pay is not that paid streaming is universally better. It is that a specific subscription removes a restriction you care about. If free services already carry enough of what you watch and their advertising does not bother you, paying may add little. If you need one exclusive catalog, dependable offline viewing, higher playback quality, more simultaneous devices, or fewer interruptions, a subscription can solve a concrete problem.

• Opera’s free built-in VPN can now switch itself on public, unsecured Wi-Fi
• The opt-in feature is rolling out to users in the US and France first
• Opera VPN Free still only protects your browser

Opera browser is making it harder to forget about your VPN.

The company’s free, built-in VPN for desktop can now turn itself on when you connect to a public or unsecured Wi-Fi network.

I would argue that Bose helped popularize clip-on open earbuds—the kind of bud that wraps around the outside of your ear like a statement earring. The Bose Ultra Open Earbuds launched in 2024 and were a big swing on the form factor, especially for a major brand. Other companies like Shokz had been making bone-conduction and open-ear headphones for years before, but this small, fashionable form factor really took off.

Bose is expanding its clip-on open earbud offerings with the new Bose Sport Open Earbuds and Bose Ultra Open Earbuds (2nd Gen), both of which launch October 1. Hence, the original Ultra Open Earbuds are on sale, matching the lowest price I’ve seen for them.

For more audio recommendations, check out our guides to the Best Wireless Earbuds, Best Noise-Canceling Earbuds, and Best Soundbars.

Love WIRED? Add us as a preferred source on Google to see more of us.

The Bose Ultra Open Earbuds are premium open earbuds that wrap around the outside of your ear, leaving your inner ear open to hear the world around you. They do not have noise-canceling of any kind. That makes them a great option for when you want to listen to music or an audiobook but still be aware, whether that’s outside going for a run or in your home washing the dishes.

In my open earbuds guide, they are my pick for the best premium open earbuds. They have better controls, sounds, and comfort than the competition, and their big con, price, is much more palatable at 33 percent off. I love them particularly for outdoor running and cycling, where I like a podcast or audiobook, but still need to hear all my surroundings.

Power up with unlimited access to WIRED. Get best-in-class reporting and exclusive subscriber content that’s too important to ignore. Subscribe Today.

Small motors inside tiny robots hit a wall the moment a machine needs a sudden shove. A UCLA and University of Michigan team built around that wall by giving a 98.2-gram platform, 11 centimeters long, 8 wide, and 4 tall, a pair of rear legs that wind up slowly and then fire all at once.

Each leg begins as a superelastic Nitinol rod, with one end secured and the other attached to a micro-servo via a pushrod. The servo twists a rod with a slight bend and begins turning it into a helix to build up elastic energy, which is stored until the rod reaches a point where it snaps into a new shape. That release happens in the blink of an eye, much faster than the servo can swing the limb on its own. When the servo spins in reverse, it resets the rod for the following cycle. It took researchers months to figure out which combinations of bend and twist would result in a lovely, clean snap rather than a sluggish, slumpy fall, and then they tweaked the helix to ensure that the burst of energy remained useful and the reset was simple.

LEGO Disney & Pixar Wall-E & EVE Building Set for Adults, Ages 18+ – Home Office, Book Shelf, or Room…

• LEGO SET FOR ADULTS – The WALL-E and EVE (43279) building set offers adults 18 years old and up an immersive construction challenge featuring…
• 4 DISNEY PIXAR CHARACTERS – Builders can create iconic robots WALL-E, EVE, M-O and Hal from the hit movie—each with authentic functionality like…
• MINDFUL BUILDING EXPERIENCE – This detailed construction set lets builders practice advanced construction techniques for an immersive and relaxing…

Two of these legs are mounted on the back of a body that is powered by a 7.4-volt LiPo battery pack and a custom control board. You can steer the thing by varying the speed of the servos. The same hardware can handle a variety of surfaces, including wood, cloth, acrylic, leather, grass, and even sand. On wood, the snap-powered robot achieved a speed of 3.21 body lengths per second. Across the six varied surfaces, it averaged 2.46 body lengths per second. However, a nearly identical model with inflexible legs was only able to reach 0.79. On cloth and grass, the rigid version came to a halt, but the snap-powered robot continued to move.

There was no trouble going up or down steps. In a sandbox filled with rocks, they even used a remote control to maneuver the robot around obstacles. Later, light sensors enabled it to navigate to a lamp by itself. With enough accumulated energy, the identical snap would send the body into repeated backflips. At this point, they sealed the circuitry and attached some thin flexible fins to the looping limbs. In water, those snaps formed paddle strokes, propelling it at a rate of nearly half a body length per second. Despite the wind pushing against the surface, the robot was still able to turn and navigate past obstructions.

Khalid Jawed from UCLA’s Structures-Computer Interaction Lab and Xiaonan Huang from Michigan Robotics spearheaded the entire project, along with co-first authors Dezhong Tong and Jiaqi Wang, as well as Zexiong Chen from Vassar College, Andy Borum from Vassar College, and Weicheng Huang from Newcastle University. It was all supported by National Science Foundation funds and published in Science Advances as Geometry-controlled instability pathway selection in elastic helices allows for fast, efficient robotic locomotion.

Because the snap is determined by the curvature of the rod rather than its size, Jawed believes the same concepts can be used to robots as small as a few millimeters in diameter. Huang emphasizes the larger point: let the structure do the work that would otherwise require more powerful motors or additional control code. For a machine small enough to vanish in tall grass, a single well-timed snap is all that separates getting stuck and jumping away from being able to paddle on in the water.
[Source]