Feed aggregator

We've always liked testing Western Digital's WD Red Pro line of NAS HDDs, which emphasize reliable for long-term storage. During reviews, we've found it difficult to argue with the resilience of this drive.

For a limited time, the WD Red Pro 18TB hard drive is $700 (was $855) at Newegg when you use code BNWF49. That equals almost 4 cents per GB. While not the cheapest high-capacity drive (especially as component costs soar), for sheer reliability, the Red Pro range comes recommended by us.

• Best for: NAS owners, home lab builders, and small businesses that need high-capacity, always-on storage.

• Why we like it: 18TB of capacity in a single drive, 7,200rpm speeds with a 512MB cache, a 300TB-per-year workload rating for NAS systems with up to 24 bays, and a five-year warranty. Other capacities are available, from 4TB to 26TB.

• What you need to know: In general, we recommend sticking to a specific brand and model for your NAS system, as they're designed to work in harmony with others.

Today's best NAS drive deal

Red Pro 18TB: was $854.99 now $699.99
A 3.5-inch 18TB NAS hard drive (model WD181KFGX) spinning at 7,200rpm with a 512MB cache and a SATA 6Gb/s interface. It’s rated for workloads of up to 300TB per year in NAS systems with up to 24 bays and carries a five-year limited warranty. View Deal

Why we recommend it

WD’s Red Pro line is a long-standing favorite among NAS users, built for the vibration and constant activity of multi-bay systems that run around the clock.

We haven’t reviewed the 18TB model, but we did test out the 14TB version in our 4-star review. It's not the fastest, but being designed for reliability over the long term is arguably more important if you're spending this much on a HDD.

For most users, 18TB strikes the sweet spot between high-capacity storage and lower initial investment compared to the 22TB to 26TB offerings.

Price Context & Historical Value

Currently, it feels like every component is spiking in price, and this HDD is no different. Just a few years ago, this same model sold for $300. Still, in today's market, it represents a good deal rather than a historic low. And the $155 off gives it a solid saving on one of WD’s highest-capacity NAS drives. The five-year warranty is a real advantage over standard desktop drives, which often come with only one or two years of coverage. If you need the space now, it’s worth grabbing before the code expires.

Should you buy it?

✅ Buy the WD Red Pro 18TB if... you need high-capacity, always-on storage for a NAS or home server, you want a drive rated for heavy workloads in multi-bay systems, or you value a five-year warranty.

❌ Skip the WD Red Pro 18TB if... you only need a simple backup drive for one PC, since an external drive is easier and cheaper; you’re building a quiet setup, since 7,200rpm drives are audible; or you can wait in case hard drive prices ease.

• See all external drives at Newegg.

Google’s new Android app store program brings one-tap installs and automatic updates, but US users are excluded, and marketplaces face a $10,000 annual fee.

The post Google’s New Android App Store Program Costs $10,000 a Year appeared first on TechRepublic.

GM agrees to a $220 million settlement over drivers' data. Learn how connected-car information was allegedly shared and what vehicle owners should know.

The post GM’s $220 Million Data Privacy Settlement: What Drivers Need to Know appeared first on TechRepublic.

Russia launched 289 attack drones, 17 ballistic missiles and 59 cruise missiles across nine regions, President Zelenskyy said in an online post that noted the barrage coincided with U.S. elections.

(Image credit: Ukrainian Emergency Service)

As polls close on Election Day and results start rolling in, these are some of the House races that will help explain how the midterms are unfolding.

(Image credit: Ryan Murphy)

Nintendo's new Ocarina of Time remake enshrines its place in the gaming canon for generations to come. But despite its stupendous production values, the game left me feeling oddly bereft.

(Image credit: Nintendo)

The CMA has published roadmaps for Big Tech before, just not for Microsoft. With £400 billion in annual public sector procurement and millions flowing to Microsoft’s cloud, the CMA can’t wait until February 2027 to act.

The topic of public procurement is becoming a major issue for UK economic policy. Sarah Cardell, the CMA's Chief Executive, delivered a speech at a Parliamentary event at the House of Commons on 8 September, introducing two reports by the CMA published that day: Public Procurement in the National Interest and Rigged Bids, Real Costs.

The headline conclusion is, perhaps unsurprisingly, that UK public procurement could be leveraged far more effectively to shape markets and improve outcomes for citizens, businesses and taxpayers alike. But a more nuanced point, and arguably the most important one, is that the biggest risks arise when short-term cost savings are prioritized at the expense of long-term value and competition.

Cloud computing is one of the starkest illustrations of this trade-off in practice. Buying Microsoft is, in the immediate term, the path of least resistance: it is already embedded in most public-sector IT estates, integration is straightforward, and the near-term cost of switching or diversifying looks far higher than the cost of simply renewing. That short-term logic is precisely why Microsoft has wound itself so tightly around the machinery of UK public-sector cloud infrastructure.

Each individual procurement decision may look like sensible, low-risk value for money at the time it is taken but cumulatively, those decisions have entrenched a single supplier so deeply into public-sector data, identity and security architecture that it now poses a very real threat to competitive procurement in public-sector AI. The apparent saving today is a major liability for tomorrow, once every AI rollout has to be built on top of infrastructure only one vendor fully controls.

A market out of balance

The CMA's cloud market investigation reached a clear verdict in July 2025: competition wasn't working well. Microsoft and AWS each hold 30–40% market share with Microsoft's share still growing, and barriers to entry including data-center costs, economies of scale, incumbent product breadth. have been deemed likely to keep it that way.

Apparently, fewer than 1 per cent of customers switch cloud providers annually, held back by egress fees and technical friction. Most tellingly, the CMA found Microsoft withholds certain licensed products from AWS and Google that directly weaken those rivals; a problem with clear implications for public-sector buyers seeking competitive procurement.

A new AI market, losing balance

However, as Microsoft rapidly rolls out new AI tools across UK enterprises and the public sector, the fundamental issue of cloud infrastructure lock-in is being overlooked. The true barrier to a competitive AI market isn't just the stickiness of new daily workflows, but the interconnected stack comprising software, data, identity, security and the foundational grip Microsoft maintains through its legacy enterprise software, specifically Windows Server and SQL Server.

Once organizations are penalized for running these essential core workloads on rival clouds, their entire data gravity shifts permanently into Microsoft's ecosystem. Building subsequent AI processes and agents around these locked-in data structures makes opting for competing, innovative AI solutions financially and technically unviable.

The CMA has already diagnosed how Microsoft uses these core licensed products to weaken rivals; they must urgently address this foundational lock-in before it permanently closes off public procurement for enterprise AI as well.

Progress, but is it too cautious?

The CMA hasn't been idle. On 31 March 2026, the CMA accepted voluntary commitments from Microsoft and AWS rather than the binding designations recommended by its own investigators. These commitments included extending Microsoft's free egress-switch window from 60 to 180 days and a new cloud interoperability request mechanism, with a six-month review promised.

It simultaneously opened a strategic market status (SMS) investigation into Microsoft's entire business-software ecosystem including Windows, Office, Teams, and Copilot, encompassing over 15 million UK users with a decision due around February 2027.

That's a fair, if cautious, step, and critics such as the Open Cloud Coalition have welcomed it while pressing for swift action if commitments aren't met. Microsoft can reasonably argue these were offered in good faith to head off heavier intervention, but we need to start seeing these promises come into fruition quickly before further entrenchment occurs. Compare an AI rollout announced in June and largely complete by October against a regulatory clock that’s usually measured in years.

A roadmap for competition and social value

The CMA is now dealing with two closely connected issues: how public procurement can deliver greater value for taxpayers and society, and how Microsoft's market position can limit competition and choice. An interim roadmap could help address both at once. It would give public bodies clearer guidance on what to consider when procuring cloud and AI services, while setting out how the CMA intends to tackle the underlying lock-in that can narrow those choices.

This would support Andy Burnham's wider focus on delivering social value in public procurement. A more competitive and open cloud market would not just benefit businesses. It could give public bodies greater flexibility, reduce the risk that short-term savings lead to expensive, irreversible dependence on one supplier, and deliver better taxpayer value at the same time.

Certainty helps everyone

The roadmap should cover non-discriminatory licensing on rival clouds; interoperability extending to identity, permissions and agent portability; public-sector contracts with built-in exit plans; and independently verifiable evidence, not just documentation, that voluntary commitments are producing real switching and multi-cloud adoption. It should also provide broader transparency about how the CMA will assess whether these measures are working in practice.

Microsoft, its customers and its competitors all have an interest in predictable rules. Microsoft would know which integration practices remain legitimate; public bodies could invest in procuring AI at scale knowing it will not close off tomorrow’s choices, and rival cloud and AI suppliers would know if building interoperable products for the UK market is commercially worthwhile.

The CMA has diagnosed the problem, taken a first cautious step, and opened the right investigation. It should now set out the destination, the milestones and the tests for success before another year of AI adoption makes Britain's cloud and workplace-software markets harder to reopen. Without a clear roadmap or tests for success, we risk anti-competitive behavior leaking from the cloud market into deeply embedded AI work environments, at a far higher cost for the UK public sector to rewind.

We've featured the best cloud database.

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

For small business owners, an unexpected payment might look like good news. A new customer has placed an order, an invoice has been settled, or an unfamiliar transaction has landed in the account. But when money arrives without an obvious explanation, the smartest response isn't to celebrate, It's to investigate.

Payment scams exploit the everyday pressures of running a business. Entrepreneurs are busy, cash flow matters, and customer relationships depend on responding quickly and professionally. Fraudsters understand this, using unexpected payments, urgent messages and convincing instructions to persuade business owners to make decisions before they've had time to check the facts.

And unfortunately, the risks are becoming harder to ignore as criminals use increasingly sophisticated digital tools to make these communications more convincing and their scams easier to scale. For small businesses without dedicated fraud teams, knowing what to look for, and having the right safeguards in place, can make the difference between a suspicious transaction and an extremely costly mistake.

Why do "random" payments happen?

An unexplained payment isn't automatically fraudulent, but it should always deserve scrutiny. Scammers can use unexpected transfers as the starting point for several different schemes, exploiting the assumption that money received is money earned.

One common tactic is the refund scam, where a fraudster sends money to a business, apparently by mistake, and then contacts the owner asking for it to be returned. They may apologise profusely, invent an urgent personal emergency or insist that a quick refund is needed to resolve an accounting error. The catch is that they ask for the money to be sent through a different channel, such as a bank transfer or gift cards.

The original payment may have come from a stolen account or compromised payment card, so if the legitimate account holder later disputes the transaction, the initial payment can be reversed, leaving the business out of pocket for the money it sent back.

Another possibility is testing stolen payment cards or hacked accounts. Criminals may use compromised payment details to make small purchases or transfers, checking whether the credentials still work. An apparently legitimate transaction could therefore be part of a wider fraud operation, and if the real cardholder disputes the payment, the business may face a chargeback, fees or the loss of goods already supplied.

Then there are phishing setups, in which the payment itself is little more than bait. Once the recipient contacts the sender, the fraudster may provide a link to a fake payment portal, request login credentials or ask the business owner to confirm sensitive information. What begins as a question about an unexplained transaction can become an attempt to take over an account.

The common overall thread is manipulation - the scammer wants the business owner to act quickly, bypass normal procedures or share information they would ordinarily keep private that they can now exploit or sell on to the highest bidder.

(Image credit: Getty Images)The warning signs every business owner should know

Although payment scams can take different forms, they often share a recognisable set of red flags. Spotting them early gives businesses the opportunity to pause before money or sensitive information changes hands.

Look out for these warning signs:

• An urgent request for a refund, where the sender insists the money must be returned immediately, perhaps citing a deadline or personal emergency.
• Pressure to use a different payment method, where the scammer asks you to return money by bank transfer, gift card or cryptocurrency instead of using the original payment platform.
• Instructions delivered outside official channels, such as unexpected emails, messages and links which direct you to alternative websites or ask you to disclose account details.
• Payments that don't make business sense. A round-number payment arrives that doesn't match an invoice, order or product price, or the sender's details don't match the customer record.

The key principle is simple, but usually effective: never assume that money received is money safely yours to return or spend. Check the transaction independently through your payment provider or bank, and never follow refund instructions supplied by an unverified sender. If a payment looks suspicious, contact the provider through its official website or app and ask how to handle it - never initiate a separate repayment yourself.

Building a stronger defence against payment fraud

Avoiding these scams doesn't require a dedicated fraud department, but it does require vigilance alongside consistent processes and a willingness to verify unusual requests rather than act on instinct.

One method is to ensure you always keep business and personal finances separate. Dedicated business accounts make transactions easier to reconcile and suspicious activity easier to identify, and mixing personal and company payments can obscure unusual transfers and potentially expose personal finances to business-related fraud.

Another useful protection is understanding what your normal transaction patterns look like. Keeping track of typical order values, customer locations, payment methods and refund volumes should be common practice - but although a transaction that falls outside these patterns isn't necessarily fraudulent, it may deserve a closer look. Reconcile payments against invoices and orders regularly, and investigate discrepancies promptly.

Securing the accounts that control your money will also go a long way. Make sure you always use strong, unique passwords for banking, payment platforms and business email accounts. Enable two-factor authentication (2FA) wherever available, and turn on alerts for large transactions, password changes and new logins, and where possible, limit account access to employees who genuinely need it.

Giving your team clear rules is another key step, as anyone who handles customer payments should know that unexpected refunds require verification. Establish a policy that prohibits returning unexplained payments through alternative channels, and encourage staff to check suspicious messages with a colleague before responding. Regular training can help employees recognise fake invoices, phishing links and attempts to impersonate customers or suppliers.

Finally, recognise that payment fraud often begins outside the payment platform itself. A fake invoice might arrive by email, a fraudulent website might imitate a trusted service, or a phishing message might steal the credentials needed to access a legitimate account. That's why good financial procedures need to be backed by effective cybersecurity.

(Image credit: Getty Images)Why Bitdefender Ultimate Small Business Security is a valuable partner

For small businesses looking to bring these safeguards together, Bitdefender Ultimate Small Business Security offers a practical way to strengthen protection without having to build an in-house security operation.

The service combines antivirus and device protection with email security, password management, VPN access, digital identity protection for all employees and business asset exposure monitoring. Its scam protection features are particularly relevant to payment fraud, helping users identify and block suspicious messages, links and websites.

Its Scam Protection Pro tool allows employees to check a suspicious message or link before responding, clicking or paying, offering an additional layer of scrutiny when an unexpected payment request or refund instruction doesn't feel right. Web and email protection can also help guard against phishing attempts designed to steal credentials or direct employees towards fraudulent sites.

Centralised management makes it easier for business owners to oversee protection across employees and devices, while support for Windows, macOS, Android and iOS reflects the reality these days that key business activity rarely takes place on a single machine or ecosystem.

Financial controls, and suspicious payments should always be verified with the relevant bank or payment provider, but combining those controls with cybersecurity tools designed to detect digital scams can make it much harder for fraudsters to exploit a moment of confusion.

For start-up owners entrepreneurs, the lesson is clear: an unexplained payment is a reason to stop and check, not an invitation to act quickly. With consistent procedures, informed employees and a comprehensive security solution such as Bitdefender Ultimate Small Business Security, small businesses can reduce their exposure to payment scams while keeping their attention where it belongs - on serving customers and growing the business.

Try our free 30-day protection built for busy founders, not IT teams

Commercial engagement in biopharma used to center largely on direct conversations between sales reps and physicians, or pharmacists. As therapeutic areas have grown more specialized, more teams have become involved in engaging with healthcare professionals (HCPs), including key account managers, medical science liaisons, brand marketers and service teams. But these functions often work separately, leaving HCPs with disjointed experiences rather than interactions that build on one another and widen their understanding.

Biopharma companies are also engaging with a much broader network of HCPs, patients, health systems and administrators. Many of these stakeholders are limiting their contact with the industry, and are increasingly using AI to inform their decisions. As clinical complexity grows and access to HCPs becomes more restricted, companies need commercial models that are more responsive to customers’ needs.

To respond to these expectations, biopharmas are turning to Agentic Commercial, connecting AI with field teams, digital channels and approved content to create more consistent, personalized HCP experiences. Built into industry-specific applications, agentic capabilities help teams capture real-world insights, uncover treatment barriers, provide trusted answers and deliver relevant content across human and AI-led interactions. This leads to more meaningful engagement, better-informed HCPs and faster access to the right medicines for patients.

Field operations that turn conversations into real-time insight

An Agentic Commercial model transforms sales teams from promotional distribution channels into strategic sources of customer intelligence, evolving CRM from a system of record into a system of guidance.

Reps conduct nuanced HCP conversations every day, but traditional reporting systems restrict insights capture to rigid picklists and drop-down menus. Critical qualitative context — the 'why' behind a physician’s decision or the localized barrier to access — goes unrecorded.

Agentic call reporting uses agentic voice and free-text capabilities, with built-in compliance checks, to help reps capture richer details from HCP conversations. These interactions generate Commercial Evidence — actionable, real-world insights that reveal intent, content gaps, and treatment barriers facing physicians and patients.

Commercial leaders gain immediate visibility into real-world friction points. Early implementations show that 65% of Agentic Call Reports identify actionable treatment barriers, enabling field teams and medical specialists to address clinical concerns with precision.

Digital experiences built around immediate answers

Field teams build trusted human relationships, but they cannot be everywhere at once. Digital channels such as brand websites and email should extend the dialogue and enable more timely responses to questions, yet too often, physicians and patients must navigate websites or lengthy documents to find what they need.

Conversational AI transforms static digital experiences into real-time conversations. Using only medical, legal, and regulatory (MLR)-approved content, agents provide trusted, compliant answers to natural language questions in real time.

These interactions also generate valuable Commercial Evidence. Instead of showing only what users clicked, AI reveals what they asked, where they struggled, and the information they could not find. Commercial teams can identify content gaps, uncover emerging treatment barriers, and trigger timely follow-up from the field.

Approved content that gives reviewers more time for higher-value work

Continuous engagement depends on a steady stream of timely, compliant content. Yet, with global pharma content expenditure reaching $30 billion annually and 29% more assets created year-over-year, content volume is outpacing human review capacity.

In an Agentic Commercial model, accelerating content speed to market is essential. Agentic MLR acts as an end-to-end expert reviewer — understanding document context, mapping specific regulations, and executing targeted agent reviews against approved labels, claim libraries, and brand SOPs. This shifts human reviewers from line-by-line proofreading to high-impact work.

By accelerating content review while maintaining compliance, commercial teams can respond faster to customer questions across field and digital channels. This delivers a more cost-effective approval process, with content leaders anticipating 38% of the MLR process to be AI-driven by 2028.

Combining AI capabilities with human expertise

Agentic Commercial brings AI and commercial teams together within the same applications, following shared processes, governance and compliance requirements. By taking on repetitive tasks and analysing large volumes of data, agents give people more time to apply their scientific expertise and judgement, shape strategy and build trusted relationships.

Bringing field teams, digital engagement, approved content and AI together allows biopharmas to create a more connected and personalized experience across channels. As therapies become more complex and access to HCPs more limited, Agentic Commercial can help teams engage customers with greater precision, support better-informed decisions and ultimately help patients access the right medicines sooner.

We've reviewed and ranked the best Electronic Health Records software.

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

• Google leads efforts to dismantle 1.3 million device residential proxy botnet built through malware preinstalled on Android hardware
• Researchers say manufacturer Videostrong intentionally shipped malware and sold botnet access to cybercriminals
• Google seized infrastructure, blocked the company from Android certification, and strengthened device security checks

Google has revealed its role in taking down a massive residential proxy network and the malware that built it - with the unusual twist that the developer was a seemingly legitimate business with multiple products and countless trusting customers.

In a blog post, the company said that together with its partners, it dismantled a backdoor called XCPE. This malware was apparently used for ad fraud, served as a dropper for additional malware, and used for assimilating devices into an illegal residential proxy network.

In total, more than 1.3 million set-top boxes alone were part of the network, the company said.

No supply chain attack

Malicious residential proxy networks are nothing new - they emerge and get dismantled on a regular basis. But this one was unique, as it came preloaded on Android devices built and sold by a certified Android Original Equipment Manufacturer (OEM) and Original Design Manufacturer (ODM). In other words, people were thinking they were buying a legitimate product from a legitimate, Google-vetted company, but were instead getting malware-laden hardware that abused their trust and sold their bandwidth to cybercriminals.

In this case, not only did the company know what it was doing, but it was actively developing the malware and even worked on hiding it from Google and everyone else.

“A device manufacturer intentionally infecting their users with malware is an unacceptable breach of trust, and represents a serious risk to those users,” Google said, not hiding its disappointment with the manufacturer.

Videostrong ended up selling two things: On the surface, it was selling Android-powered smartphones and similar devices that contained malware that built a botnet. Under the table, it was selling access to that botnet to cybercriminals.

Google was joined by a number of other companies and agencies in this operation, including Japan’s National Police Agency, Lumen, HUMAN Security, Plume, Synthient, and TrendAI.

Threat actors are always in need of residential proxy botnets. They can use them to launch powerful Distributed Denial of Service (DDoS) attacks, hide their addresses as they operate, conduct large-scale credential stuffing campaigns, bypass geographical restrictions, scrape websites at scale, and more. We don’t know how much Videostrong made from selling access to the botnet, or who used it.

Taking down the C2 domains

To disrupt both the malware and the network, Google took down the domains and servers the company used to control the botnet and install malware modules, took down C2 domains that were used by over a dozen malware families loaded as modules via the XCPE backdoor, and banned Videostrong and all future subsidiaries and affiliates from building and selling Android devices in the future, or supplying software to other certified vendors.

Google also said it had improved its firmware integrity checks for certified Android TV devices to prevent similar incidents involving other manufacturers.

“Google requires OEMs go through a comprehensive suite of compatibility and security tests to ensure a consistent and safe user experience across the Android ecosystem,” it stressed. “However, Videostrong employed obfuscation techniques to exploit a gap in our tests and load malicious code after the device's first boot. As a result, Google has implemented changes to its existing review process to better safeguard against similar activity in the future.”

This is not the first time an electronic device came with malware off the shelf. Last month, security researchers Bitdefender found Android phones from multiple manufacturers, including Doogee and Cubot (both Chinese firms) being sold with preloaded malware. However, in those cases, the supply chain was compromised and the malware was included without the OEM’s knowledge, or consent.

Google's AI satellite generated a poem from orbit. Learn how Project Suncatcher could shape the future of space-based data centers and AI computing.

The post Google’s First Space Data Center Satellite Goes Online and Writes Poetry appeared first on TechRepublic.

OpenAI chief scientist Jakub Pachocki seemed to issue a warning on September 6.

In “An Alien Mind,” he argued that current alignment and monitoring techniques may soon prove inadequate for the pace at which machine intelligence is advancing. He also described a future in which AI takes a growing role in improving AI, compressing the time available for human supervision and institutional response.

Much of AI safety research concentrates on behavior. For example, what an AI system is permitted to do, how its actions are monitored and where humans retain approval. Yet every action begins one step earlier, with a judgment about what information deserves belief.

That earlier step is incredibly important.

AI acts on information whose pedigree is often obscure

Before an AI sends a payment, rewrites software or recommends a decision, it first assembles a picture of the world. That picture can draw on search results, corporate records, research papers and material produced by other AI systems.

The web was built for circulation. Information travels easily, while its ancestry often fades as it passes from one publication to another. A claim can begin in a research paper, appear in a news story, be condensed into a blog post and surface months later in a forum thread. By the time an AI encounters the statement, the route back to the original evidence may be difficult to reconstruct.

Authority also gets compressed. A peer-reviewed study, a corporate announcement and an anonymous post can all arrive as text. An AI system can rank and combine them into a polished answer, while the reader sees little of the difference between documented evidence, advocacy and repetition. Disagreement adds a further complication.

Credible sources regularly reach different conclusions, especially in finance, medicine, politics and emerging technology. AI is usually rewarded for producing a useful answer, which creates pressure to reconcile conflict into a single sentence. Uncertainty becomes easy to erase precisely when it deserves to remain visible.

The consequences grew substantially once AI acquired the ability to act. An agent can place a trade, approve a purchase, alter software or send a message on someone’s behalf. An uncertain claim can therefore become the premise for an irreversible action.

Human oversight depends on better evidence

The effectiveness of human oversight depends on the reviewer having superior information or a reliable means of checking what the AI used. In many settings, the human reviewer has access to the same web that the AI searched, only at a fraction of the speed. The AI can scan thousands of pages while the person responsible for approval works through a summary and a handful of links. As capability rises, review risks becoming a procedural signature rather than an independent examination.

Pachocki writes that OpenAI’s ability to rely on chain-of-thought monitoring is diminishing as reasoning becomes more complex and as AI becomes more capable of manipulating its own reasoning process. He calls for safety thresholds that can be enforced by third-party auditors, government agencies or international bodies.

Those safeguards govern capability and behavior. Reliable oversight also requires an evidentiary record showing what information produced the conclusion, where each claim originated, how trustworthy the source is and where credible disagreement remains. Without that record, the reviewer is judging the answer more than the evidence behind it.

AI inherited the web’s weakest habit

AI absorbed an information culture built around speed, reach and replication. Search engines, social feeds, aggregators and reposts made access extraordinarily efficient. They also encouraged a world in which the same statement could appear hundreds of times while tracing back to a single uncertain source.

Generative AI intensifies that effect because AI systems now produce material that later AI systems can ingest. A weak claim can be generated, scraped, summarized and republished until repetition gives it the appearance of consensus.

This creates a provenance problem with compounding consequences. Every additional generation can separate a statement further from the evidence that gave rise to it. Once an autonomous agent treats that statement as reliable, the information chain becomes part of the decision chain. While for consumers, this can mean a poor recommendation, for a financial institution, insurer or government agency, the same weakness can affect money, access, security or public services.

Autonomous AI needs a system of provenance

Take finance. Serious financial activity depends on records that establish who acted, under whose authority, on which terms and with what audit trail. Institutions spend heavily on those records because trust requires more than a plausible outcome. Autonomous AI needs a comparable discipline for information.

Claims should carry provenance that survives copying and transformation. A person or an agent should be able to trace a statement back to its origin, inspect the evidence behind it and see how the statement changed as it travelled. Credible disagreement should remain intact.

Where reputable sources diverge, the system should preserve the competing views and the degree of confidence behind each one. For an autonomous agent, uncertainty can serve as a trigger for further research, human approval or a narrower course of action.

Accountability should travel with information as well. A claim from an identifiable institution with supporting evidence deserves different treatment from an unattributed assertion repeated across dozens of pages. The aim is to give AI a richer sense of evidentiary quality before it commits resources or acts on someone’s behalf.

Governance belongs at a shared level. Provenance standards, verification methods and accountability rules need independent scrutiny and participation from many institutions, rather than dependence on the preferences of a single provider.

Final thoughts

Pachocki argues that current alignment and monitoring techniques may soon constrain the pace at which labs can responsibly scale AI tools, and he expects voluntary slowdowns until shared safety thresholds emerge. He also warns that continued progress could carry AI into recursive self-improvement, with AI driving a larger share of its own development.

That leaves every company building autonomous systems with a fundamental task. Establish what the agent knows before it acts and make that knowledge auditable.

Behavioral safeguards remain essential. So do permissions, audits and human approval. Their value rises sharply when the information beneath every decision carries provenance, context and accountability. The next phase of AI safety, therefore, begins with the evidence an agent sees, the authority it assigns to that evidence and the uncertainty it preserves before taking action.

Autonomy built on verifiable information gives humans a stronger basis for control. Opaque information leaves supervisors judging decisions whose foundations remain difficult to inspect.

Pachocki is right to warn that the window for preparation may be narrow. The industry should use that window to build systems that can show where their knowledge came from before those systems are trusted to act on the world.

We've featured the best AI chatbot for business.

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

Google confirms a Pixel Watch battery drain issue after owners report severe power loss. Here's which models users mention and what to do while waiting for a fix.

The post Google Confirms Pixel Watch Battery Drain Issue, Promises Fix appeared first on TechRepublic.

The rules have led to changes in course materials, prompted accusations of censorship and fueled pushback from faculty and students, including faculty departures, student protests and a federal lawsuit.

(Image credit: Michelle Waida)

NPR investigated cyberattacks on the digital devices that control industrial refrigerators, revealing the extent to which operational technology is exposed to damage and disruption.

(Image credit: Fred Tanneau)

A massacre allegedly targeting Hindu Americans was recently thwarted, but researchers say it helps to illustrate a growing and increasingly mainstream hate directed at South Asians in the U.S.

(Image credit: Laura Bargfeld)

Race calls are a critical part of election night and NPR partners with the Associated Press to make those calls. Here's how we do it.

(Image credit: Justin Sullivan)

One last check in with the Swing Shift voters before Election Day. All of them voted for President Trump. Now, they're rating his presidency and deciding who to support when he's not on the ballot.

(Image credit: Tara Anand for NPR)

There are a lot of variables for how things will go today. Results will start to come in at 6 p.m. ET in Kentucky and Indiana, and then it will be off to the races.

(Image credit: Jon Cherry)

Compare Linear vs Jira for the best agile project management tool and explore their pricing, Agile features, AI capabilities, and integrations.

The post Linear vs Jira (2026): Which Is Better for Software & Product Teams? appeared first on TechRepublic.