Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More The era of reasoning AI is well underway. After OpenAI once again kickstarted an AI revolution with its o1 reasoning model introduced back in September 2024 — which takes longer to answer questions but with the payoff of higher performance, especially on complex, multi-step problems in math and science — the commercial AI field has been flooded with copycats and competitors. There’s DeepSeek’s R1, Google Gemini 2 Flash Thinking, and just today, LlamaV-o1, all of which seek to offer similar built-in “reasoning” to OpenAI’s new o1 and upcoming o3 model families. These models engage in “chain-of-thought” (CoT) prompting — or “self-prompting” — forcing them to reflect on their analysis midstream, double back, check over their own work and ultimately arrive at a better answer than just shooting it out of their embeddings as fast as possible, as other large language models (LLMs) do. Yet the high cost of o1 and o1-mini ($15.00/1M input tokens vs. $1.25/1M input tokens for GPT-4o on OpenAI’s API) has caused some to balk at the supposed performance gains. Is it really worth paying 12X as much as the typical, state-of-the-art LLM? As it turns out, there are a growing number of converts — but the key to unlocking reasoning models’ true value may lie in the user prompting them differently. Shawn Wang (founder of AI news service Smol) featured on his Substack over the weekend a guest post from Ben Hylak, the former Apple Inc., interface designer for visionOS (which powers the Vision Pro spatial computing headset) and co-founder of Dawn, an analytics and diagnostics platform for AI products. The post has gone viral, as it convincingly explains how Hylak prompts OpenAI’s o1 model to receive incredibly valuable outputs (for him). In short, instead of the human user writing prompts for the o1 model, they should think about writing “briefs,” or more detailed explanations that include lots of context up-front about what the user wants the model to output, who the user is and what format in which they want the model to output information for them. As Hylak writes on Substack: With most models, we’ve been trained to tell the model how we want it to answer us. e.g. ‘You are an expert software engineer. Think slowly and carefully“ This is the opposite of how I’ve found success with o1. I don’t instruct it on the how — only the what. Then let o1 take over and plan and resolve its own steps. This is what the autonomous reasoning is for, and can actually be much faster than if you were to manually review and chat as the “human in the loop”. Hylak also includes a great annotated screenshot of an example prompt for o1 that produced a useful results for a list of hikes: This blog post was so helpful, OpenAI’s own president and co-founder Greg Brockman re-shared it on his X account with the message: “o1 is a different kind of model. Great performance requires using it in a new way relative to standard chat models.” I tried it myself on my recurring quest to learn to speak fluent Spanish and here was the result, for those curious. Perhaps not as impressive as Hylak’s well-constructed prompt and response, but definitely showing strong potential. Separately, even when it comes to non-reasoning LLMs such as Claude 3.5 Sonnet, there may be room for regular users to improve their prompting to get better, less constrained results. As Louis Arge, former Teton.ai engineer and current creator of neuromodulation device openFUS, wrote on X, “one trick i’ve discovered is that LLMs trust their own prompts more than my prompts,” and provided an example of how he convinced Claude to be “less of a coward” by first “trigger[ing] a fight” with him over its outputs. All of which goes to show that prompt engineering remains a valuable skill as the AI era wears on. source

Each year, we calculate how much business growth improving Forrester’s Customer Experience Index (CX Index™) by one point drives. For 2024, we published the results in the report, How Customer Experience Drives Business Growth, 2024. The report includes the dollar upside of improving CX Index by one point for 12 industries: airlines, luxury auto manufacturers, mass-market auto manufacturers, auto/home insurers, multichannel banks, direct banks, credit card issuers, health insurers, midscale hotels, upscale hotels, investment firms, and retailers. A Sneak Peek Into The Business Growth From CX In 2024 The benefits of improving CX can be massive. For example, for a mass-market auto manufacturer, improving CX by one point can lead to more than $1 billion in additional revenue — this is because improving CX increases the chance that customers will buy their next car from the same brand and take the car to the brand’s dealership for service needs. For an auto/home insurer, it’s close to $370 million. In many industries, the upside of making a happy customer even happier is higher than that of placating an unhappy customer. This is because the growth benefits of improving CX increase exponentially when going from “good” to “excellent” for those industries, which include some financial services industries. CX pros in firms where this relationship holds true must focus on identifying CX drivers that move customers from “OK” and “good” CX scores to “excellent” scores. The effect of recommendations on the business upside of CX is small. For each of the industries in our analysis, acquiring new customers via recommendations accounts for less than 7% of the overall business benefit from improved CX. Calculate These Numbers For Your Own Firm Should you use our numbers to communicate the value of CX in your firm? Yes and no. Use them to get initial buy-in that CX drives business results. But don’t just assume that your numbers will look the same. Instead, calculate the business upside of CX for your own firm. Here is how we calculated it — hopefully, you will find this useful: 1. Calculate what each customer is worth, depending on how loyal they plan to be. Forrester’s Customer Experience Benchmark Survey measures the quality of customers’ experiences and their loyalty intentions. Together with other data, we calculate a revenue potential for each customer. How we calculate it depends on how companies in each industry make money. 2. Create models that link CX Index and revenue potential. Our analysis shows, for each industry, the effect of CX changes on business outcomes. We also found out whether that effect changes based on whether we go from a low CX Index score to a medium one or from a medium one to a high one. 3. Calculate the upside of improving CX by one point. Our model shows the impact on a customer’s revenue potential when the CX Index score of the industry rises by one point. We then multiplied that per-customer upside by the number of customers of a big brand in the industry (we focused on the largest brands in the CX Index in each industry, as a few big brands dominate each industry that we investigate). Forrester Clients: Use Our Five-Step Solution Blueprint To Calculate The Business Impact Of CX The image below shows step one. Click Prove That CX Efforts Produced Business Results for all details.   Thank you for your major contribution to this research, James Williams! source

  越捷航空再度由著名航空安全和產品評級網站AirlineRatings認可，榮獲2025年低成本航空公司中「全球最安全航空公司」名銜。 越捷與其他全球領導低成本航空公司，如瑞安航空 （Ryanair）、易捷航空（easyJet）和邊疆航空（Frontier Airlines）一同躋身前十名。越捷多年來致力提供舒適安全服務，是次認可反映航空公司對飛行安全的承諾，為乘客和機組人員提供安心保障。 在2025年AirlineRatings評估中共有385家全球航空公司，當中考慮過去兩年的事故記錄、機隊年齡、國際民航組織（ICAO）審核結果以及其他安全措施等因素。 AirlineRatings.com首席執行官Sharon Petersen表示：「越捷機隊不但是越南最年輕，更擁有亞太地區最新飛機，多年來安全記反映了其卓越成就。」AirlineRatings還為越捷頒發了七星安全評級，是當中最高的評級，此評級自2018年以來頒發至今。 目前，越捷航空擁有現代和高燃油效率的空中巴士機隊，技術可靠性達到99.7%，與全球名列前茅。同時，越捷繼續在培訓、維護和工程方面進行大量充足投資，以維持最高的安全標準。越捷航空學院（VJAA）作為國際航空運輸協會（IATA）的培訓夥伴，確保持續供應高技能的航空專業人才。 越捷現時營運３條航線飛往富國島、峴港及胡志明市。為滿足2025年的旅遊高峰期，來往香港至峴港及富國島航線已提升為每日運營。越捷是首家從香港直飛富國的航空公司，香港市民可享有 30 天免簽證待遇。 香港至越南的航班 往返香港至富國島   香港→富國島 VJ985 16:40 – 18:25  (2小時45分鐘) 每日一班 富國島→香港 VJ986 11:55 – 15:40  (2小時45分鐘) 往返香港至峴港 香港→峴港 VJ967 21:20 – 22:10  (1小時50分鐘) 每日一班 峴港→香港 VJ966 17:35 –20:20  (1小時45分鐘) 往返香港至胡志明市 香港→胡志明市 VJ877 19:50 – 21:30 (2小時40分鐘) 逢星期一、三、五、日 胡志明市→香港 VJ876 15:10 – 18:50 (2小時40分鐘)   LinkedIn Email Facebook Twitter WhatsApp The post 越捷航空再度獲AirlineRatings認可 榮獲2025年全球最安全航空公司之一 appeared first on VeriMedia. source

CyberGhost VPN fast facts Our rating: 4.3 stars out of 5Pricing: Starts at $6.99 per month (6-month plan)Key features: Offers servers in 100 countries and 125 locations. Includes streaming, torrenting, and gaming servers. Has generous free trial options. CyberGhost VPN has an impressive server fleet spaning 100 countries and 125 locations, an affordable starting price, and specialized servers for streaming, torrenting, and gaming. It also supports Windows, macOS, Linux, iOS, Android, Android and Apple TV, and even gaming consoles. While its complicated company history may cause some concern, the overall VPN package you get with CyberGhost makes it a worthy VPN choice of the year. Semperis Employees per Company Size Micro (0-49), Small (50-249), Medium (250-999), Large (1,000-4,999), Enterprise (5,000+) Large (1,000-4,999 Employees), Enterprise (5,000+ Employees) Large, Enterprise Features Advanced Attacks Detection, Advanced Automation, Anywhere Recovery, and more ESET PROTECT Advanced Employees per Company Size Micro (0-49), Small (50-249), Medium (250-999), Large (1,000-4,999), Enterprise (5,000+) Any Company Size Any Company Size Features Advanced Threat Defense, Full Disk Encryption , Modern Endpoint Protection, and more ManageEngine Log360 Employees per Company Size Micro (0-49), Small (50-249), Medium (250-999), Large (1,000-4,999), Enterprise (5,000+) Micro (0-49 Employees), Small (50-249 Employees), Medium (250-999 Employees), Large (1,000-4,999 Employees), Enterprise (5,000+ Employees) Micro, Small, Medium, Large, Enterprise Features Activity Monitoring, Blacklisting, Dashboard, and more CyberGhost VPN pricing Plans Price One month $12.99 per month (14-day money-back guarantee) Six months $6.99 per month (45-day money-back guarantee) Two years $2.03 per month (45-day money-back guarantee) While some VPNs offer different features at varied price points, CyberGhost’s pricing is straightforward. It provides three prices depending on your preferred service length — either one month, six months, or two years. The monthly option is billed every month, while the six-month option is billed every six months at $41.94 and the two-year option is billed for the first two years at $56.94 then annually thereafter. In terms of affordability, you will get the most out of your money with their two-year plan. At $2.03 per month, CyberGhost is one of the most affordable VPN options out right now. This is especially true once we take into account the number of servers you get at this price. Keep in mind, however, that this contract locks you in for two years so you won’t have the flexibility of being able to switch providers quickly. If you’re looking to test out the VPN for a month, CyberGhost VPN’s $12.99 is on the pricier end compared to its competitors like Proton VPN, with its one-month Plus plan currently priced at $9.99 per month. You also don’t get CyberGhost VPN’s in-house NoSpy servers for free with the one-month plan. With that, I do wish that CyberGhost offered at least a one-year option for its paid plans instead of its six-month subscription. This would give consumers more leeway with cost and less of the time commitment. Annual plans are also standard offerings for most VPN providers. In my view, having a one-year option would help CyberGhost provide a better one-to-one pitch against its competition. Comparing it to other VPNs, Surfshark’s mid-tier One subscription for one year is priced at $3.39 per month, while NordVPN’s Plus plan is $5.49 per month. At $6.99 per month for only six months, I feel like CyberGhost may seem like less of a bargain when looking at what other VPNs have to offer. CyberGhost VPN does have a fairly generous free trial program. For Windows and macOS users, you can get a 24-hour free trial without providing any payment or credit card information. Android users also have access to a three-day trial, while iOS users get free seven-day access. While the 24-hour Windows and Mac trial is short, accessing all the premium features without handing off any financial information is a worthwhile trade-off. Is CyberGhost VPN safe? CyberGhost VPN offers all the industry-standard security protocols we expect from a decent VPN in 2023. It has the OpenVPN, IKEv2, and WireGuard protocols, as well as AES-256 encryption. It also has a kill switch, DNS leak protection, and split tunneling capability. On its website, CyberGhost boasts that its headquarters are based in Romania — a country with strong privacy laws. This is a big plus for users who are wary of government interference. However, it is worth noting that CyberGhost is owned by Kape Technologies, formerly Crossrider. Crossrider had formerly been associated with malware and adware controversies before it purchased CyberGhost back in 2017. While CyberGhost has maintained that it acts as an independent entity of Kape, the company history is important to mention when it comes to security and privacy. CyberGhost does have a no-logs policy, which states that they don’t log user data such as browsing history, IP addresses, session durations, and the like. This no-logs policy has been confirmed by an independent audit conducted by Deloitte in 2022. In April 2024, CyberGhost completed a second independent audit of its server network and management systems. The assessment was conducted by Deloitte Audit Romania, which said that it found no evidence to conclude that CyberGhost’s “configuration of IT systems and management of the supporting IT operations” were not in accordance with its No-logs Policy No-Logging Safeguards. I commend CyberGhost for continuing its commitment to undergoing third-party audits. I especially appreciate how recent their latest audit was, providing an additional layer of confidence to users interested in purchasing a CyberGhost subscription today. The VPN service also regularly publishes quarterly Transparency Reports that cover police requests and malicious activity reports they receive. This is an admirable level of transparency that definitely adds to CyberGhost VPN’s overall credibility. With its complicated history, prospective users may be hesitant to consider CyberGhost VPN at first. However, I do feel that the company is doing enough to alleviate security concerns customers may have. Key features of CyberGhost VPN CyberGhost VPN provides an alluring package of features for its price, making it a solid choice. Let’s take a look at some key features one by one. Extensive server network CyberGhost

00:00 100:00:03,966 –> 00:00:06,000Welcome to CIO Leadership Live.200:00:06,000 –> 00:00:10,133I’m Lee Rennick, executive directorof CIO communities for cio.com,300:00:10,366 –> 00:00:13,233and I’m very excited and honored todayto welcome Christine400:00:13,233 –> 00:00:16,500Bongard,CEO, co-founder of the WIT Network.500:00:16,800 –> 00:00:17,800Christine.600:00:17,800 –> 00:00:20,500Please introduce yourselfand welcome to the show.700:00:20,500 –> 00:00:20,933Hi there.800:00:20,933 –> 00:00:22,833Lee, thank you so much for having me ontoday.900:00:22,833 –> 00:00:26,500Yes. Christine Bongard,CEO and co-founder of the WIT network.1000:00:27,266 –> 00:00:30,000I have been a woman in technologymy whole career.1100:00:30,000 –> 00:00:33,933I had the fortunate, fortunate experienceof being part of a Techstars1200:00:34,033 –> 00:00:38,866up in the New York metro areaand, had the opportunity to grow business1300:00:38,866 –> 00:00:42,300from start and learn all about everything1400:00:42,333 –> 00:00:45,600technology, how to manage teams,how to run a business.1500:00:45,600 –> 00:00:49,100It was just such an incredible opportunityfor me1600:00:49,400 –> 00:00:52,500and, thatthat led me to being a woman in tech.1700:00:52,833 –> 00:00:57,733And I got to build a great network,over all of that time and,1800:00:58,933 –> 00:00:59,733have the fortunate1900:00:59,733 –> 00:01:03,600experience to then take my, careeras an entrepreneur2000:01:03,600 –> 00:01:08,100and in the tech industry and builda global network for women in tech.2100:01:08,466 –> 00:01:10,366as my my second part of my career.2200:01:10,366 –> 00:01:11,700So loving it.2300:01:11,700 –> 00:01:14,366Wow. Wow, that sounds really amazing.2400:01:14,366 –> 00:01:16,000And we’ll dive into thata little bit more.2500:01:16,000 –> 00:01:18,000And, you know, I’ma member of the Witt network.2600:01:18,000 –> 00:01:21,166Our organization is, I’m so proud that2700:01:21,166 –> 00:01:24,166we have that connectionwith this organization.2800:01:24,166 –> 00:01:26,433and all thatyou do to ensure that women have a safe2900:01:26,433 –> 00:01:29,433place,a place to network and learn and share.3000:01:29,433 –> 00:01:32,133And, so I’m looking forwardto having this interview today.3100:01:32,133 –> 00:01:35,200So I really do appreciate youjoining us today.3200:01:35,200 –> 00:01:36,266Christine, thanks so much.3300:01:36,266 –> 00:01:39,800We have created this seriesto support diversity in technology3400:01:40,000 –> 00:01:43,000and really listen to womenworking in this space.3500:01:43,066 –> 00:01:45,900and how, you know, they’re buildingand supporting out,3600:01:45,900 –> 00:01:48,833other women in, in, in our industry.3700:01:48,833 –> 00:01:51,833And so the first questionand I ask everyone this question,3800:01:52,233 –> 00:01:55,566could you please tell us a little bitabout your own career path and maybe3900:01:55,566 –> 00:01:59,166provide some insights or tips on that rolepath, especially as a women woman?4000:01:59,166 –> 00:02:02,133Excuse me, are there any lessons sharedthat you could, you know,4100:02:02,133 –> 00:02:03,900any lessons learned that you could share?4200:02:03,900 –> 00:02:05,500So a couple of things.4300:02:05,500 –> 00:02:05,966You know, as4400:02:05,966 –> 00:02:10,466I mentioned, just a great opportunityto come up in tech throughout my career.4500:02:10,466 –> 00:02:13,866And I will say thatjust about the whole way through,4600:02:14,300 –> 00:02:18,500I was, consistently one of the only womenin the room, whether we were4700:02:18,500 –> 00:02:23,000with a customer, a partner,at a conference, at networking events,4800:02:23,933 –> 00:02:26,933you know, leads to feeling isolated.4900:02:27,066 –> 00:02:31,733and, you know, some of those eventswere golf events and whiskey tastings5000:02:31,733 –> 00:02:36,666and cigar, events, and those were thingsI really couldn’t relate to.5100:02:36,666 –> 00:02:39,300And, you know,it was hard to break into conversations5200:02:39,300 –> 00:02:42,166with people and I thought,where were the women?5300:02:42,166 –> 00:02:43,200You know, we have to5400:02:43,200 –> 00:02:46,766the technology industry is suchan incredible place to be.5500:02:47,166 –> 00:02:51,066And, you know, just felt like,how do we bring more women into this5600:02:51,066 –> 00:02:54,766and, and make, you know,bring gender parity to these events?5700:02:55,066 –> 00:02:59,000so that you mentioned the word safe,you know, so that that women feel5800:02:59,000 –> 00:03:03,500more comfortable, you know, being in,being in the room and being present.5900:03:03,500 –> 00:03:07,666So, you know,I really took those experiences that6000:03:07,666 –> 00:03:11,566that fueled, my need to, to want to bea part of starting this network.6100:03:11,966 –> 00:03:14,533But you asked about, you know,what are some lessons learned?6200:03:14,533 –> 00:03:16,366And I would say, you know, given6300:03:16,366 –> 00:03:20,166that we haven’t seen a ton of changein the numbers for gender parity,6400:03:20,433 –> 00:03:24,566in this industry,you have to really be bold and go for it.6500:03:24,566 –> 00:03:29,133And, you know, I think, you know,sitting back and kind of waiting for6600:03:29,366 –> 00:03:33,266career paths to materializein front of you or for your manager6700:03:33,533 –> 00:03:38,666to take the time to help you build it out,or for even HR teams to, to do that.6800:03:38,666 –> 00:03:39,966We’re we’re in a,6900:03:39,966 –> 00:03:43,133we’re in a place now in the worldwhere they just don’t have the time7000:03:43,133 –> 00:03:44,233to do that stuff. Right.7100:03:44,233 –> 00:03:49,233So I think we have to really inspire womento get clear on their vision.7200:03:49,300 –> 00:03:50,300Where do they want to go?7300:03:50,300 –> 00:03:54,333What skills do they want to build,and then look at the networks around them,7400:03:54,333 –> 00:03:58,800look at, you know, look at wherethey can take advantage of learning.7500:03:59,133 –> 00:04:02,433You know, we’ve got lots of stuffon LinkedIn learning, and there’s lots of,7600:04:02,633 –> 00:04:07,300technology companies and organizationslike the Witt network that provide7700:04:07,500 –> 00:04:11,466skills training and different,you know, educational opportunities.7800:04:11,466 –> 00:04:14,566And so I just really encourage womento be bold.7900:04:14,566 –> 00:04:15,400Go for it.8000:04:15,400 –> 00:04:18,400You know, get out there and and capture,8100:04:18,533 –> 00:04:21,733you know, the skills that they needto help accelerate their careers.8200:04:22,500 –> 00:04:24,066Yeah, I think that’s great advice.8300:04:24,066 –> 00:04:27,600I was just speaking to a friendand colleague about just this very thing.8400:04:27,866 –> 00:04:31,666I have my own personal like networkingbank book that I have on It’s8500:04:31,700 –> 00:04:35,100a rainbow and it’s on my desk,and it’s like I look at it and when I8600:04:35,100 –> 00:04:39,266need those moments to kind of plan and,you know, speak to others.8700:04:39,266 –> 00:04:42,833And I have a business coach I work with,you know, that’s super important.8800:04:42,833 –> 00:04:46,800And I didn’t really learn this until, likenow in life or like, you know, when I,8900:04:47,066 –> 00:04:50,666when I, started working with organizationslike the IT network.9000:04:50,666 –> 00:04:52,933So, it’s really impactful to me.9100:04:52,933 –> 00:04:56,400So maybe we could, we couldthe segue as well into the next question.9200:04:56,700 –> 00:04:59,933could you please tell our audiencea little bit about the Witt network,9300:05:00,333 –> 00:05:01,566you know, what is it?9400:05:01,566 –> 00:05:02,700How is it formed?9500:05:02,700 –> 00:05:06,300And, you know, we talked about thisa little bit, but why it’s important right9600:05:06,300 –> 00:05:09,300now, especially for women in tech,to have organizations like.9700:05:09,600 –> 00:05:12,700Oh, gosh, I could talk about this for,for

The Federal Communications Commission is putting the word out about a new scheme aimed at tricking people into thinking their homes will be foreclosed on unless they make emergency payments into an account controlled by the scammers. source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Pharmaceutical giant GSK is pushing the boundaries of what generative AI can achieve in healthcare areas like scientific literature review, genomic analysis and drug discovery. But it faces a persistent problem with hallucinations, or when AI models generate incorrect or fabricated information. Errors in healthcare are not merely inconvenient; they can have life-altering consequences. Here’s how GSK is tackling it. The hallucination problem in generative health care A lot of focus around reducing hallucinations has been applied during the training of a large language model (LLM), or when it is learning from data. But to mitigate hallucinations, GSK instead employs strategies at inference-time, or at the time when a model is actually being used in a real application. Strategies here include things like self-reflection mechanisms, multi-model sampling and iterative output evaluation. According to Kim Branson, SVP of AI and machine learning (ML) at GSK, these techniques help ensure that agents are “robust and reliable,” while enabling scientists to generate actionable insights more quickly. “We’re all about increasing the iteration cycles at GSK — how we think faster,” he said. Leveraging test-time compute scaling Improving an generative AI application’s performance at inference-time, also referred to as test-time, is mostly done by increasing computational resources when a model trying to figure out the answer to a problem. This includes more complex operations such as iterative output refinement or multi-model aggregation, which are critical for reducing hallucinations and improving model performance. Branson emphasized the transformative role of scaling this phase of test-time compute in GSK’s AI efforts, noting that by using strategies like self-reflection and ensemble modeling, GSK can leverage these additional compute cycles to produce results that are not only quicker, but more accurate and reliable. In fact, this is a broader industry trend, not only across healthcare, but other verticals too. “You’re seeing this war happening with how much I can serve, my cost per token and time per token,” said Branson. “That allows people to bring these different algorithmic strategies which were before not technically feasible, and that also will drive the kind of deployment and adoption of agents.” Strategies for reducing hallucinations To tackle hallucinations in healthcare gen AI apps, GSK employs two main strategies that require additional computational resources during inference. Self-reflection and iterative output review One core technique is self-reflection, where LLMs critique or edit their own responses to improve quality. The model “thinks step by step,” analyzing its initial output, pinpointing weaknesses and revising answers as needed. GSK’s literature search tool exemplifies this: It collects data from internal repositories and an LLM’s memory, then re-evaluates its findings through self-criticism to uncover inconsistencies.  This iterative process results in clearer, more detailed final answers. Branson underscored the value of self-criticism, saying: “If you can only afford to do one thing, do that.” Refining its own logic before delivering results allows the system to produce insights that align with healthcare’s strict standards. Multi-model sampling GSK’s second strategy relies on multiple LLMs or different configurations of a single model to cross-verify outputs. In practice, the system might run the same query at various temperature settings to generate diverse answers, employ fine-tuned versions of the same model specializing in particular domains or call on entirely separate models trained on distinct datasets. Comparing and contrasting these outputs helps confirm the most consistent or convergent conclusions. “You can get that effect of having different orthogonal ways to come to the same conclusion,” said Branson. Although this approach requires more computational power, it reduces hallucinations and boosts confidence in the final answer — an essential benefit in high-stakes healthcare environments. The inference wars GSK’s strategies depend on infrastructure that can handle significantly heavier computational loads. In what Branson calls “inference wars,” AI infrastructure companies — such as Cerebras, Groq and SambaNova — compete to deliver hardware breakthroughs that enhance token throughput, lower latency and reduce costs per token.  Specialized chips and architectures enable complex inferencing routines, including multi-model sampling and iterative self-reflection, at scale. Cerebras’ technology, for example, processes thousands of tokens per second, allowing advanced techniques to work in real-world scenarios. “You’re seeing the results of these innovations directly impacting how we can deploy generative models effectively in healthcare,” Branson noted.  This week, in a partnership with Mayo Clinic and Microsoft, Cerebras announced a genomic foundation model that predicts the best medical treatments for people with rheumatoid arthritis using the efficiencies found in its custom silicon. When hardware keeps pace with software demands, solutions emerge to maintain accuracy and efficiency. Challenges remain Even with these advancements, scaling compute resources presents obstacles. Longer inference times can slow workflows, especially if clinicians or researchers need prompt results. This is where the advanced silicon comes in. Higher compute usage also drives up costs, requiring careful resource management. Nonetheless, GSK considers these trade-offs necessary for stronger reliability and richer functionality.  “As we enable more tools in the agent ecosystem, the system becomes more useful for people, and you end up with increased compute usage,” Branson noted. Balancing performance, costs and system capabilities allows GSK to maintain a practical yet forward-looking strategy. What’s next? GSK plans to keep refining its AI-driven healthcare solutions with test-time compute scaling as a top priority. The combination of self-reflection, multi-model sampling and robust infrastructure helps to ensure that generative models meet the rigorous demands of clinical environments.  This approach also serves as a road map for other organizations, illustrating how to reconcile accuracy, efficiency and scalability. Maintaining a leading edge in compute innovations and sophisticated inference techniques not only addresses current challenges, but also lays the groundwork for breakthroughs in drug discovery, patient care and beyond. This is part of our Healthcare and Gen AI feature series. source

The U.K. government is considering banning ransomware payments to make critical industries “unattractive targets for criminals.” It would apply to all public sector bodies and critical national infrastructure, which includes NHS trusts, schools, local councils, and data centres. Currently, all government departments nationwide are banned from paying cyber criminals to decrypt their data or prevent it from being leaked. This rule intends to protect the services and infrastructure the British public relies on from financial and operational disruption. The health sector is classified as CNI, so withholding ransomware payments could impact patient care. According to Bloomberg, the attack on pathology company Synnovis last June, which led to months of NHS disruption, resulted in harm to dozens of patients, with long-term or permanent damage in at least two cases. SEE: Number of Active Ransomware Groups Highest on Record Organisations must also report ransomware attacks within three days On top of the ban, the proposed legislation will make it mandatory for organisations to report ransomware attacks within 72 hours of becoming aware of it. This is so law enforcement remains up-to-date on whom is being targeted and how which aids their investigations into organised crime groups and enables them to publish helpful advisories. The Home Office also wants to instate a ransomware payment prevention regime involving educating businesses on responding to a live threat and criminalising unreported payments. It is hoped that this will both increase the National Crime Agency’s awareness of attacks and reduce the number of payouts made to hackers, especially in exchange for data suppression. On Jan. 14, the Home Office opened a consultation on these three proposals, which will run until April 8. Ultimately, the goal is to reduce the sum of cash criminals extract from U.K. companies and boost understanding of the ever-changing ransomware landscape to aid prevention and disruption efforts. “These proposals help us meet the scale of the ransomware threat, hitting these criminal networks in their wallets and cutting off the key financial pipeline they rely upon to operate,” security minister Dan Jarvis said in a press release. The proposed approach to improving the country’s cyber security appears to echo that of the U.S. The federal government mandates compliance with its cyber security initiatives for federal agencies and regulated industries, hoping other businesses will voluntarily follow suit. Must-read security coverage Blanket ban could disproportionately impact small businesses and non-critical sectors Within the documentation outlining the proposals, the Home Office acknowledges the potential for the legislation to disproportionately impact small and micro-businesses “which cannot afford specialist ransomware insurance, or clean up specialists.” These SMBs will have less employee capacity during an attack to engage with the government and meet reporting deadlines. As a result, they may feel that the only option to retain their business is to pay to decrypt data. SEE: 94% of Ransomware Victims Have Their Backups Targeted Alejandro Rivas Vasquez, the global head of Digital Forensics and Incident Response at security firm NCC Group, said in a statement that the blanket rule could create “unfair and administrative burdens that become complex and unmanageable” for smaller businesses. He said: “Instead of a one size fits all approach, we’d recommend the government explore a less burdensome obligation that could be applied to smaller businesses, or focus on incentivising businesses to improve their security posture, rather than punitive action.” Vasquez added that applying the ban only to public sector bodies and CNI could impact other industries. “A blanket ban could place a larger target on sectors not included in the ban, such as manufacturing, which doesn’t currently fall under the scope,” he said. Manufacturing was the second most targeted industry for ransomware last year, after services, and saw a 71% year-on-year increase. Furthermore, the legislation would not impact hackers who are motivated by factors other than money. As Vasquez said: “In geopolitically motivated attacks, which can be launched by nation states, ransomware is a tool to cripple critical national infrastructure and steal sensitive data – money is not the objective. Banning payments would be futile in stemming such attacks – the hackers would already have the data they need.” U.K.’s cyber risks are ‘widely underestimated’ In December, Richard Horne, head of the U.K.’s National Cyber Security Centre, warned that the country’s cyber risks are “widely underestimated.” He said that hostile activity had “increased in frequency, sophistication, and intensity,” largely from foreign actors in Russia and China. According to the NCSC’s Annual Review 2024, the agency handled 430 incidents this year compared to 371 in 2023. Of these, 13 were “nationally significant” ransomware incidents threatening essential services or the wider economy. SEE: Microsoft: Ransomware Attacks Growing More Dangerous The report called ransomware the most pervasive threat to U.K. businesses, especially in academia, manufacturing, IT, legal, charities, and construction. According to the NCSC, the pervasion of generative AI has been found to increase the risk of ransomware by providing “capability uplift” to attackers. Amateur attackers can use it to craft social engineering materials, analyse exfiltrated data, code, and reconnaissance, which essentially lowers the barrier to entry. source

Coming from South Africa but living in Europe, I can tell you that sending money to family and friends back home is a bit of a nightmare. Typically you must use a traditional bank, which can take a week or more, or payment apps like PayPal or Wise, which charge high fees.  The antiquated nature of remittance payments is something that immigrants are all too familiar with. Demand for better alternatives is giving rise to a new cohort of fintech companies looking to streamline the process. One of them is London-headquartered LemFi. Founded in 2021, the financial services platform enables diaspora communities in North America and Europe to quickly and affordably send money to friends and family in China, India, Pakistan, Nigeria, Kenya, and 15 other countries in the Global South. Sadly, for me, South Africa is not yet on the list. However, LemFi is expanding fast so I might not have very long to wait. LemFi has already onboarded 1 million customers so far, who have made a combined $1bn in monthly transactions through the app. Transactions to and from Asia are currently growing at 30% month-on-month, said the company. And last week, LemFi, which employs over 300 people, officially set up shop in Europe. The startup is tapping a global remittance market predicted to reach $1.3 trillion by 2032.   The 💜 of EU tech The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now! All that growth potential, has, unsurprisingly, piqued the interest of VCs. Today, LemFi announced that it has raised $53mn in Series B funding. London-based growth-stage venture firm Highland Europe led the round, with participation from previous investors Left Lane Capital, Palm Drive Capital, and Y-Combinator. The fresh funding brings LemFi’s total raised to $85mn. “When we started building LemFi, we were told remittance had already been solved,” said Ridwan Olalere, co-founder and CEO of LemFi, pictured left. “But for too many people, it is still too slow, cumbersome and expensive with customers telling us that in some instances it was cheaper to send money from the US via Canada than directly to their families back home.” Olalere, originally from Nigeria, founded LemFi alongside Norwegian Rian Cochran. The pair met whilst working at Nigerian fintech unicorn OPay, incubated by Norwegian browser provider Opera.  Fuelled by fresh funding, LemFi now looks to add new features and expand to new countries, said Olalere. No doubt part of LemFi’s popularity is that it advertises zero transaction fees. Well, unless you live in China, India, or Pakistan. The company also makes revenue on foreign currency exchanges. Its business model is dependent on volume, making small profits on numerous transactions while staying competitive for users seeking low-cost international money transfers. LemFi is far from alone in the increasingly crowded remittance fintech market. Perhaps the most well-known is US-based Remitly, which went public in 2021. Other contenders include startups Zepz and Taptap Send, both of which are based in London.      source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Scams are everywhere. McAfee’s new scam detector spots and stops scams across text, email, and video to keep you from being fooled. McAfee today announced at CES 2025 the launch of McAfee Scam Detector – the most comprehensive protection against text, email, and video scams. Today’s scams are smarter, sneakier, and more convincing than ever – and they’re everywhere. One in three Americans admit falling for a text, email, or video scam in the last 12 months. From fake emails and suspicious texts to deepfake videos that look incredibly real, scammers are using clever tricks to steal people’s money and personal information. McAfee is helping consumers take back control with its AI-powered McAfee Scam Detector to stop scammers in their tracks. McAfee is using AI to try to stop scams before they strike via automatic detection of scams or potential scams. Scammers have a couple of things working in their favor, said Steve Grobman, CTO of McAfee, in an interview with GamesBeat. One is there’s a large number of private communication channels where there’s not necessarily a moderator, like on encrypted messaging or direct messaging. They could be communicating with the victim, you know, under the guise of security. The other piece is real time, and it’s one of the reasons that McAfee has made its deepfake detector product work on any video stream in the web, Grobman said. Beginning this spring, McAfee’s Scam Detector will be included at no extra cost for McAfee customers. This must-have product uses the latest in advanced AI technology to proactively analyze and flag risky messages in real time. That ‘Hi, how are you?’ text from a stranger? It’s one of the top text scams of 2024. I got a scam email about using an automated document signing service, with an attachment saying “termination NDA” on it. I almost fell for that one, given the urgency I associated with that one. (I wasn’t fired). An urgent email about a failed delivery? Probably fake. And no, Elon Musk doesn’t have a unique investment opportunity for you. McAfee makes it easy to tell real from fake in seconds and gives you the winning combination of tips and technology to keep scams out of your life for good. “I think the thing that is most clever is the personalization,” said Grobman. “One of the things that we have seen are job scams, where people have made their search public, and those scams take the form of taking your job interview to the next level, but it requires a background check and we require the applicant to pay for the background check.” He said the scanning for scams is working in real time, making it possible to catch the scammers at the right times. Background A year ago, Grobman talked to me about focusing on scam defense in three areas: scams that come through email, scams that come through text, and then scams that users are exposed to, more generally, in other media forms like video. To address that, the company developed advanced AI models to protect consumers in all three of those areas and it started to deploy them in late 2024. One of those was a deepfake detector to help consumers identify whether AI generated videos are AI generated versus authentic. The company also ran technology previews for our other modalities, namely email and text. “And now that we’re moving into 2025 we’re very excited to move all of these technologies into a much larger scale by making them available to the vast majority of our customers,” Grobman said. “We’re evolving our AI models to take advantage of advanced AI PCs when an advanced AI PC is there. So we can run our AI models on the [PC’s own] NPU, but we also have the ability to run on other inference engines, either the CPU or GPU, which is able to give us a broader scope, to give capabilities to a broader set of users.” The email scam defense is going to move beyond a technology preview that McAfee did in 2024 where it was available for Microsoft-based email properties. Now it can support Gmail and other properties. “In offerings in 2025, we’ve developed advanced technology to detect deep fake images. So we did a partnership with Yahoo News in 2024 around helping to ensure that images that came through the news pipeline, that their editors, their quality assurance personnel, could best detect if anything was generated with AI image generation technology and help provide that insight.” Protecting people in today’s Scamiverse Every day, scammers trick people with fake emails, texts, and videos, and the results can be devastating. Americans report receiving 12 scam messages daily, losing as much as $1,000 when they fall for one, and spend 80 hours a year simply trying to figure out if the onslaught of messages they receive are real or fake, according to a McAfee survey. And deepfake scams, which use AI to create fake video, can be even worse – some people have lost up to half a million dollars, based on a McAfee survey. It’s clear that scams have become a drain on people’s time, energy, and finances. “I’ve had more issues with scam texts recently. I’d say within the last year, it’s just been bad. Like I’ve been getting a lot of spam emails, texts, calls…it’s a lot,” said Tina, 31, in South Carolina. “It was a fake email for UPS. I thought that I was signing up to change my address and instead it charged my credit card,” said Alexandria, 46, in Georgia. “While I love new technology, I have been very scared of AI generated videos and information since seeing just how realistic it can be,” said Haley, 24, from New Jersey. “Scammers are getting smarter every day, using technology like artificial intelligence to make their tricks more convincing and harder to spot,” said