You’re receiving this email because you subscribed to the newsletter. If you’d like to unsubscribe, you can manage your preferences here with one click (or use the button at the bottom of the email).

Welcome to this edition of Loop!

To kick off your week, I’ve rounded-up the most important technology and AI updates that you should know about.

‏‏‎ ‎ HIGHLIGHTS ‏‏‎ ‎

Let's jump in!

1. “Vibe-hacking” is now a top AI threat

We start this week with a report from Anthropic, which outlines how AI is being used by sophisticated criminals. The company recently disrupted a cybercrime group, who were using Claude Code to extort data from 17 organisations - with this all being done in a matter of weeks.

The victims included healthcare providers to government entities, with ransom demands exceeding $500,000. Unfortunately, AI is being used to accelerate these types of attacks.

Previously, these types of operations would have required entire teams of people. But with agentic systems that can use different tools, that’s no longer the case and a single person can now co-ordinate these attacks.

That makes it even harder for authorities to detect and track these people down, as there are fewer online messages to trace between people. According to Anthropic, their AI model essentially ran these operations "end-to-end”.

Claude was able to create extortion messages that were specifically tailored for the organisation, then calculated how much the data is worth on the dark web.

Perhaps more unsettling is the North Korean scheme uncovered. Workers with minimal coding skills or English proficiency were able to land Fortune 500 tech jobs, with Claude used to ace the interviews and maintain their positions.

Claude is also being used in romance scams, as criminals say it is a "high EQ model" and performs better than rivals. It’s worrying news for us all, as the barrier to sophisticated cybercrime has never been lower.

2. Taco Bell reconsiders its AI drive-throughs

After deploying the technology to over 500 locations and seeing mixed results, the company is now reconsidering its strategy for AI drive-throughs.

It follows some rather embarrassing viral moments, including when one customer ordered 18,000 cups of water - just for a human employee to override the system and manually complete the order.

Instead of replacing workers entirely, Taco Bell is favouring a hybrid model - with restaurant teams receiving guidance on when to rely on AI voice assistants and when to intervene. If a restaurant has lengthy queues, they recommend that human staff are used at peak times.

This is a notable shift for the industry, which became thrilled at the prospect of replacing staff and having fully-automated ordering systems.

It perfectly captures just how hard it is to bridge that gap between proof-of-concept and real-world success, especially when your AI system might suddenly think someone actually wants 18,000 cups of water.

3. Nuclear fusion is seeing rapid growth, as investors pile in

The nuclear fusion sector is experiencing remarkable growth, with companies racing to crack what's considered to be the ultimate prize in clean energy.

While conventional nuclear plants work by splitting atoms apart, fusion does the opposite - it smashes atomic nuclei together, similar to the Sun.

This could result in massive amounts of energy being created, without having to produce greenhouse gases or the radioactive waste - which is a huge problem with today’s nuclear power.

The field hit a watershed moment in 2022, when researchers managed to get more energy out than they put in - which was a first for the sector. Now everyone's scrambling to replicate and commercialise that breakthrough.

According to the Fusion Industry Association, 53 fusion companies have attracted $8.9 billion in private funding and $795 million in public support. That's a staggering leap from 2021's $1.9 billion total, across just 23 companies.

As I’ve written about before, tech giants are also betting on fusion power to support their data centres and AI initiatives. Earlier in the week, Nvidia, Google, and Bill Gates announced investment in Commonwealth Fusion Systems - which has raised a whopping $863 million.

4. US investment in clean tech slows, as cancellations outweigh new projects

America's clean tech manufacturing sector is running into trouble, with more projects being cancelled than announced in the second quarter.

Companies have pulled the plug on $5 billion worth of projects, while they announced just $4 billion in fresh investment - according to analysis from the Rhodium Group and MIT.

It’s a sharp departure from the manufacturing boom that we’ve experienced in recent years, following the signing of the Inflation Reduction Act.

Even though battery factories were hit hardest by the second quarter pullback, they still attracted the most new investment at $8 billion.

This wider shift follows on from the “Big Beautiful Bill", which led to tax credits vanishing overnight and left many planned facilities in doubt.

We’ll need to see if this trend continues in the next few quarters, but it's a big shift for a sector that was supposed to be the foundation of America's manufacturing comeback.

5. IBM and AMD will partner on quantum computing

Both companies have announced that they are joining forces to create hybrid quantum-AI computing systems.

The partnership will combine IBM's quantum expertise with AMD's AI-optimised silicon to develop commercial quantum systems. It's a fascinating bet on what could be the next major computing paradigm shift.

While rivals have led the Generative AI sector - such as Nvidia’s complete domination of the AI chip market and Microsoft’s strategic partnership with OpenAI - IBM and AMD are instead trying to get a head start on the next advancement.

The collaboration aims to make quantum computing more accessible to researchers and developers, moving beyond today's experimental systems and towards practical applications.

What's particularly interesting here is the open-source angle. Rather than building another walled garden, they're betting that an open platform will drive adoption and establish their technology as the de facto standard.

Time will tell if this pays off for IBM and AMD, but open-source has benefitted plenty of other companies - including Facebook with React, Microsoft with VS Code, and even Red Hat with Linux.

Google’s new AI model can edit photos in seconds

Earlier in the week, Google announced a new model that can edit photos with incredible precision.

As you might have already seen, generative AI models struggle to edit only one thing and will often make several changes to the image instead. With Gemini 2.5 Flash Image, Google seems to have made huge progress.

You can simply upload an image and ask the model to “make the shirt red”, with changes only made to that part of the image. It’s surprisingly good and definitely worth a try.

If you look really closely at their examples, you’ll notice that it can add a waxy-texture to the person’s skin, but overall it does an incredible job.

I can see this tool being used by property realtors, who already use basic image generators to “restyle” a room and help people to visualise how it would look with modern furniture.

Those images can look quite uncanny and clearly stand out as fake, but Google’s new image editor could completely change that and allow realtors to easily re-design the space.

This could also become a new challenger for Adobe Photoshop, which is incredibly expensive for what it is. I don’t expect professionals to suddenly ditch that tool, but ordinary users might decide it’s no longer worth the price - especially since Google charges just $0.039 for an image edit.

Of course, this AI model could also be abused by bad actors and be used to spread misinformation. To counter this, Google will add an invisible watermark to all edited images - which can be used to flag an image as AI-generated.

As I’ve covered before, Google’s SynthID watermark is widely considered as the industry’s best and is incredibly hard to defeat - as the watermark is embedded within the image itself.

📈 Nvidia posts record-breaking sales as AI demand surges

🤖 Anthropic launches a Claude AI agent that works in your Chrome browser

💔 Cracks are forming in Meta's partnership with Scale AI

🏢 Microsoft AI launches its first in-house models

📱 South Korea bans smartphones in the classroom

🎤 Will Smith is accused of using AI to create fake concert fans

💡 Mark Cuban shares his formula for disrupting industries, from entertainment to healthcare

💰 Meta will spend tens of millions on a pro-AI political campaign group

💬 Spotify introduces a chat function as it pushes into social networking

🔓 Elon Musk announces that xAI has made Grok 2.5 open source

This startup is based in Estonia and has developed an app that helps people to overcome fears with public speaking.

It’s able to coach people throughout and offers useful tips, along with challenges to better train your voice - which can range from tongue twisters to breathing exercises.

The app has cleverly positioned itself as the introverted professional's dream. It’s a safe space to practice your public speaking, without having to do it in front of actual humans.

Recent numbers suggest they're onto something: $12 million in ARR with 50,000 paying subscribers, which has been achieved on remarkably lean funding. Overall, the team has reached 4 million downloads and now has 160,000 active users.

This all started years ago, after the CEO was bullied in school for how his voice sounded. He then met a voice coach and created his own YouTube channel to help others, which eventually morphed into Vocal Image.

The company has just raised $3.6 million to help expand the development team and focus on more languages. Given the huge user base, Vocal Image has used that audience to label over 1 million voice samples and give itself an advantage over its rivals.

If you want to learn more about the startup and try the app for yourself, I’ve included a link below.

This Week’s Art

We’ve covered quite a bit this week, including:

If you found something interesting in this week’s edition, please feel free to share this newsletter with your colleagues.

Or if you’re interested in chatting with me about the above, simply reply to this email and I’ll get back to you.

Have a good week!

Liam

Share with Others

If you found something interesting in this week’s edition, feel free to share this newsletter with your colleagues.

Liam McCormick is a Senior AI Engineer and works within Kainos' Innovation team. He identifies business value in emerging technologies, implements them, and then shares these insights with others.