Connect with us

Tech

Viral ChatGPT shows first signs of risks when AI takes over and runs with it

Published

on

A PC screen featuring the home page of viral AI app ChatGBT
ChatGPT: viral AI tool has taken the world by storm, but inaccuracies include confusion about its own subscription fees | Photo: Jonathan Kemper

Unless you have been living under the same rock as those claiming to have not seen the rather peculiar attire singer Sam Smith wore at the 2023 Brit Awards, last Saturday, chances are you have heard of ChatGPT. The chatbot, launched in 2020 with the promise of using artificial intelligence to generate human-like text based on user inputs, is already the trending topic of the year. From curious students to event planners and marketing professionals, everyone wants to try and see for themselves if the content written by an AI tool can live up to its hype. So much so that, according to a recent survey by OpenAI, the San Francisco-based research lab behind ChatGPT, the tool is now being used by over 10 million people globally, with new users joining the platform every day – an intake speed that led to the number of monthly active users increasing by over 200% in the past year alone.


“We currently use ChatGPT as a quick and handy assistant for tasks such as idea generation, solving mathematical equations, helping with Excel formulas, and many other day-to-day tasks that can be automated.”, says George Bates, SEO Manager at digital agency Limelight Digital.

“Whilst it’s important to use AI with caution, we embrace it as a tool to help us work more efficiently rather than simply dismissing it as ‘unethical’ or ‘lazy’. However, we don’t yet use AI for any content yet, which seems to be the most prevalent usage at the moment,” acknowledges the UK-based professional.

Podcaster Nate Runkel has also been exploring the AI tool and is happy with the results.

“I ran two tests that I was completely impressed by. One, I plugged in some information and had ChatGPT generate the script for my podcast intro. And then to see how it flowed, I read it in my voice to my fiancée who to this day does not believe that I did not
write it. The second test I ran was to generate a press release for the upcoming season of the same podcast. And again, I was completely mesmerized not only by how it was able to create usable content, but also how much I could finesse it to match my own writing style/voice,” says Runkel who even used the tool recently to pen a cover letter for a job application.

“I think it’s more accurate than you would expect something to be that is just pulling data points it has been fed but, that being said, it still has a long way to go. And anything I’ve used I have just used as the framework and then have had to make revisions to personalize and clean up any inaccuracies!”

At Tally Workplace, co-founder Laura Beales is another convert to the AI tool that’s changing the way people write and research.

“We use chatGPT frequently within our business and believe as a small startup it can help us get ahead of the competition that is slower to make use of the tool. It greatly reduces the time needed to create content about each of our workspaces and offices. Whilst we edit the descriptions and articles it produces, we estimate it reduces the time for us to create content by 80%,” explains the entrepreneur.


In the UK, London-based PR and branding agency This is Run currently uses ChatGPT to help brainstorm ideas.

“I also use it for writer’s block and content ideas; our last three blogs on our website have all been created using AI. It frees up time and ensures we have some SEO-worthy content flowing,” says Run’s co-founder, Samuel Dontoh.

Thousands of miles from the capital of England, in Canberra, Australia, co-founder and CEO of fitness and leisure brand Salti, Jo Flynn, has also started to use AI for her business.

“Our virtual assistant/copywriter was creating all our social and email content. However, she had to leave us to focus on her studies. I haven’t been able to replace her so started experimenting with ChatGPT and I’m happy with the outcome so far. I’m a trained comms and marketing professional (with nearly two decades of experience), but I simply don’t have time these days to write content. So, ChatGPT was a simple (and free) solution,” Admits Flynn.


“ChatGPT has written this blog post about The rise of sweat to swim activewear. I requested a certain style and tone of voice, and after a few rounds landed on the above. It came up with some pretty cool stuff each time I requested a change in tone or style. Then, I made some slight edits to make sure it was accurate and reflected our brand, and that was the end result,” celebrates the entrepreneur that utilised ChatGPT to stay one step ahead.

“Afterwards, I asked ChatGPT to write a series of Facebook ads and headlines and I’m now running ads to a cold audience, direct to this blog post. Each link click costing around $0.14.”

The viral effect that has put artificial intelligence in the mainstream daily conversation has catapulted the tool to the top of the most valued startups in recent years. In 2019, even before the launch of ChatGPT less than three years ago, OpenAI had already secured a $1 billion investment from Microsoft. Last month the Wall Street Journal reported that the research lab started talks to sell existing shares in a tender offer that would value the company at around $29 billion – a significant jump in valuation from a prior tender offer completed in 2021, when OpenAI was estimated to be worth $14 billion.

Not everyone, though, has fallen head over heels for ChatGPT.

“ChatGPT is a useful tool if used correctly. Thinking that you will get a perfect copy on your first try every time is unrealistic. In general, it is best practice to use the copy that the tool creates as a first draft. It can be helpful to spark ideas and save some time in the initial stages of developing copy,” says Paula Napolitano, product sales manager at Wisetek. “Pieces generated can also have a robotic quality to them. The lack of nuance and tone gives the impression that a robot wrote the piece, meaning that you lose a personal connection with your audience. ChatGPT is by no means the end of professional copywriting. The platform still doesn’t express itself in a convincing enough way to make it believably human,” highlights Napolitano who has extensive experience in copy development.


“I use the AI tool to help me write emails, blog posts, and Twitter threads. It’s useful for outlining and idea generation, but it can’t give you a finished piece” says Marketing Consultant Michael Comeau. “ChatGPT-written content is often inaccurate, and it’s always boring. It reads like something out of a content farm. So, you have to rewrite everything it outputs, especially since Google and other search engine seem primed to downrank AI-generated content,” analyzes Comeau.

“ChatGPT responses are not 100% accurate but they are easy to curate to your needs. One plus point of the tool is that it generates very ‘accurate data in numbers’. For example, if I need to see the active number of users on Facebook, ChatGPT gives me a very accurate figure leaving room for any potential changes. For content, undoubtedly ChatGPT is a great help, but it can often wrong you. Several times I have had to curate the generated content because it is not always according to our defined personas. Sometimes it even generates totally opposite content to what we are aiming for,” explains Ben Miller, founder of Focus On Digital, who has reduced his Google search for data in numbers after starting to use ChatGPT for marketing purposes.

“ChatGPT sometimes generates totally opposite content to what we are aiming for” – Ben Miller, founder of Focus On Digital.

Some concerns arise from the limited capability of ChatGPT in understanding the sensitive layers of some content.

“We are using ChatGPT to give us the foundation for new articles we write for both our sites. However, we would never dream of simply copy/pasting this content as we know it gets flagged by Google’s algorithms and so we will use a combination of paragraph rewriting tools and our own human editors to improve the content. We have also found that it doesn’t always provide accurate information, especially when it comes to things like LGBTQ+ rights,” points out Darren Burn, CEO at Out Of Office, a full-service luxury travel specialist in inclusive LGBTQI+ travel.

“I do believe it’ll be a good tool for research for content writers, but I would urge management to ensure that its habits don’t seep into copy and content on websites just to try to get more traffic to a site. You can imagine Google is working hard to combat AI-generated SEO written content – and rightly so,” points out Burn.

And British Website Consultant Victoria Bennett, who uses the AI tool to help create content for websites, is using the tool with caution because of some other drawbacks.

“When drafting a blog article, I ask ChatGPT some questions that will help me gain background information quickly. It is accurate most of the time. But I personally wouldn’t rely on it completely and I still check key facts, as it gets some factual information wrong. Recently, while looking for information on golf courses for a blog post, ChatGPT presented me with some incorrect information on who designed a particular course. I had to check the course owner’s website to find the accurate information,” recalls Bennett.


“ChatGPT has also become a victim of its own success when it comes to availability. The tool’s popularity means that the system can become overwhelmed at times. When demand is high, the tool may not be able to admit additional users, causing temporary disruptions in service. This can be frustrating for users, especially for those who are relying on ChatGPT for critical tasks,” says Natasha Maddock from Events Made Simple.

Despite occasional challenges, Maddock believes ChatGPT remains a valuable tool for businesses and individuals looking to improve efficiency and productivity: “Time is precious. We use ChatGPT for a variety of tasks, including editing content, suggesting article topics, writing blog post outlines, and generating FAQs. We were pleasantly surprised by its capabilities,” says the professional, who spent the first 14 years of her career as a technology business adviser.

Inaccuracies delivered by ChatGPT also include out-of-date information about its own products.

Launched early in February 2023 as an alternative for those unsuccessfully trying to login into an overwhelmed website, often informing users that the platform is working at capacity and suggesting account holders try to log in another time – or leave an email to be informed when the free version of the AI tool will be up and running – ChatGPT Plus is a paid version offering advanced features and priority access at peak times in countries such as the United States, Canada, the United Kingdom, and Australia. The service costs $20 per month. However, 10 days after its launch, when asked to write a paragraph about the premium service and its benefits, ChatGPT returned my prompt informing me that it costs $49 to subscribe for ChatGPT Plus – yes, it isn’t sure even how much OpenAI charges for the usage of its own invention.

There is a reason for that.

Although ChatGPT is a clear breakthrough and has seen tremendous growth in popularity, attracting millions of users from around the world thirsty for a flexible tool for natural language processing, the wide range of content it offers currently stumbles upon another crucial limitation: time.

As the AI tool has only been trained on data up to 2021, it means that ChatGPT has no knowledge of current events. But this won’t stop the artificial intelligence tool from trying to give you some content, if asked – even if it is completely untrue. When asked to write a short article about who is the President of Brazil and when he took office, ChatGPT wrote that former military officer Jair Bolsonaro is the current President of Brazil, having taken office on January 1, 2019. In fact, Luis Inácio Lula da Silva is the actual head of the largest Latin American country.

The viral tool of 2023 also wouldn’t know that singer Sam Smith made headlines with his choice of attire for the red carpet of the Brit Awards this year: a black latex zip-up outfit featuring inflated shoulders and thighs which created a heart-shaped effect on the lower half – an image that ChatGPT couldn’t ever unsee, had it been fed with recent data.

Marcio Delgado is a Journalist, Producer and Influencer Marketing Manager working with brands and publications in Europe, America and Asia.

Tech

WhatsApp upgrades group chats with new tools

Published

on

WhatsApp upgrades group chats with new tools
Earlier this year, WhatsApp upgraded group chats with the launch of group message history, member tags and event reminders. | Photo: Julian Christian

Meta-owned messaging platform WhatsApp has announced a series of new updates designed to improve how users communicate in group chats, introducing enhanced polls, a new @all mention feature and the ability to create new chats from existing groups.

The latest additions build on several group chat improvements introduced earlier this year, as WhatsApp continues to refine one of its most widely used features. The updates aim to help users make decisions more quickly, improve communication in busy conversations and make it easier to organise discussions without creating unnecessary clutter.

New tools to simplify group decisions

WhatsApp has expanded its polling feature with three new options designed to make group decision-making more efficient.

Users can now set a closing time for polls, ensuring that voting automatically ends at a specified deadline. Poll creators also have the option to hide participants’ names, allowing members to vote privately. In addition, poll questions can now be edited within 15 minutes of being posted, making it possible to correct mistakes or clarify wording without creating a new poll.

The changes are intended to make organising meetings, events and group activities quicker and more straightforward.

New ways to keep everyone informed

Among the most significant additions is the introduction of @all, a feature that allows users to notify every member of a group with a single mention, rather than tagging individuals separately. The feature is designed for announcements such as schedule changes, event reminders or urgent updates. In groups with more than 32 members, only administrators will be able to use @all.

WhatsApp says users will continue to have control over their notifications and can choose to mute @all alerts through the app’s notification settings if they prefer. The platform has also introduced a feature that allows users to create a new group chat directly from an existing one. Instead of adding participants individually, members of an existing group can be selected with a single tap to start a separate conversation.

The latest updates follow the introduction earlier this year of group message history, member tags and event reminders, as WhatsApp continues to expand the tools available for managing group conversations.

Continue Reading

Tech

TikTok Cracks Down on AI Spam as It Expands User Education

Published

on

TikTok Cracks Down on AI Spam as It Expands User Education
Recent tests improved detection systems to target accounts dedicated to posting AI-generated spam. | Photo: Ron Lach

TikTok has announced a new set of initiatives focused on cracking down AI spam and how social media users can spot it. With more than 1 billion people using the app each month in 2025, the ByteDance‑owned company has become one of the world’s biggest public stages — a place where entertainment, commentary, activism and misinformation all travel at the same speed. And as AI‑generated content floods social feeds everywhere, TikTok is trying to stop its own platform from becoming a playground for synthetic junk.

The company’s latest update focuses on two fronts: helping users understand AI better, and stopping AI‑generated spam from overwhelming the real creators who keep the app alive.

AI literacy with focus on manipulated content

TikTok is expanding its AI literacy push with new resources created alongside NAMLE and AI and deepfakes expert Henry Ajder. The aim is to give users something more useful than vague safety tips — practical explanations of how AI tools work, how to spot manipulated content, and how to use AI without accidentally contributing to the mess.

It’s also doubling down on expert‑led content. Through the NoFiltr programme, launched in November 2025, helped trusted organisations to generate more than 200 million views with content that enables people better understand and use AI. TikTok says it has already invested more than $4 million behind the initiative and plans to keep scaling it.

“TikTok’s early and scaled implementation of Content Credentials demonstrates how provenance can deliver meaningful value in real-world environments around the world,” said Clement Wolf, Chair of the C2PA. “Bringing that experience to the Steering Committee will help accelerate our collective work to make transparency a consistent and trusted part of the digital ecosystem”.

AI spam is getting smarter

AI creativity is exploding on TikTok, but so is AI‑generated spam. The platform removed over 86 million fake accounts in the first three months of this year — a number that shows just how aggressively spam networks try to hijack trending topics.

Now TikTok is testing new detection systems aimed specifically at accounts that mass‑produce AI‑generated content on sensitive subjects like politics, current events, financial advice and medical claims.

Transparency is becoming a requirement

Two years ago, TikTok became the first major video platform to adopt C2PA Content Credentials, a technology that helps people see when a video has been generated or heavily edited using AI. It’s now joining the C2PA Steering Committee, giving it a seat at the table as transparency standards evolve across the industry.

So far, TikTok has labelled over 3 billion videos as AI‑generated using a mix of Content Credentials, creator‑applied labels and invisible watermarking.

Alongside safety measures, TikTok is still building creative AI tools like Smart Split and AI Outline, and testing features such as Manage Topics, which let users control how much AI‑generated content appears in their feed.

Continue Reading

Tech

Visa Uses AI Agents to Automate Purchases

Published

on

Visa Uses AI Agents to Automate Purchases
AI agents are now completing purchases with participating merchants. | Photo: Yoco Photography

At the Visa Payments Forum (VPF), which took place in Paris earlier this month, bringing together payment industry leaders to explore the future of payments, Visa has announced the execution of live agentic commerce transactions across Europe, with AI agents carrying out purchases at participating merchant websites on behalf of cardholders. The transactions are now taking place in live environments, marking an advance beyond testing at controlled storefronts, with AI agents now transacting based on cardholder instructions directly with independent merchants.

The global payment company says that these transactions were carried with AI agents being used to browse products, select items and initiating purchases, acting within consumer-defined parameters. Visa connects banks, merchants and AI systems via its network to enable secure, authenticated agentic transactions in line with European regulatory requirements.

Selected Merchants Join Visa’s AI Trial

Visa has and continues to support the execution of live agentic commerce transactions in Europe, working with over 30 European issuers to enable AI agents to make purchases on behalf of cardholders, with participating merchants including lastminute.com, Frasers, Cleverbridge and BrickDepot.

All the AI-enabled transactions span multiple sectors including travel, retail and e-commerce, demonstrating how agentic commerce can begin to operate across different real‑world consumer use cases.

“We’re now seeing AI agents buy on behalf of people directly with independent merchants,” said Mathieu Altwegg, Head of Product and Solutions for Visa in Europe. “The next step is to scale this by bringing the whole ecosystem together – from standards and infrastructure to partners and enablers – with trust built in from the start. It’s the same approach we took to scale contactless, and it’s how this next wave of commerce will take shape.”

Merchants participate for the first time

Merchants’ participation was enabled through Visa’s Trusted Agent Protocol (TAP) and Agent Directory – designed to help merchants to securely recognise and work with verified AI agents across different platforms and environments. These capabilities provide a consistent, secure signal of agent identity, allowing merchants to distinguish trusted AI interactions from non‑verified traffic, while maintaining control over how agents access their sites, surface products, and complete transactions.

Rather than requiring new infrastructure, these signals can be integrated into existing risk, policy and user experience frameworks. TAP is designed to work alongside both existing and emerging commerce protocols, allowing merchants to adopt it in a way that fits their preferred platforms and underlying infrastructure without disruption.

Banks Completing Live AI‑Executed Transactions

The issuing banks who have completed live, agent‑executed transactions at participating merchants – authenticated using Visa Payment Passkeys in support of issuer’s compliance with Strong Customer Authentication requirements – include: Barclays, BBVA, CaixaBank, HSBC UK, ING, Klarna, and Revolut, amongst other companies.

Visa is aiming to scale and expand agentic commerce across sectors and markets in Europe, with the model also being extended to support commercial and B2B payment scenarios.

Continue Reading

Trending